Get live statistics and analysis of Alexia Jolicoeur-Martineau's profile on X / Twitter
Senior AI Researcher at the Samsung SAIT AI Lab 🐱💻
I build generative AI for images, videos, text, tabular data, weights, molecules, and video games.
1kfollowing19kfollowers
The Innovator
Alexia Jolicoeur-Martineau is a trailblazing Senior AI Researcher passionately crafting cutting-edge generative AI models across a dazzling array of data types—from images and text to molecules and video games. With a sharp mind and a knack for defying expectations, Alexia champions ultra-efficient neural networks that challenge the status quo. This profile is a powerhouse of AI innovation, blending deep technical knowledge with a drive to push boundaries.
Alexia’s obsession with tiny neural nets shows she’s basically an AI minimalist—why build a mansion when you can just stick to an ultra-efficient shoebox stacked with layers? The real question is, does she measure success in parameter count or how many coffee cups she’s consumed during coding marathons?
Achieving state-of-the-art results on complex AI benchmarks with a neural network less than 10 million parameters is a monumental breakthrough, demonstrating profound mastery and innovation that challenges the norms of AI model scaling.
To revolutionize artificial intelligence by developing groundbreaking, efficient, and versatile generative models that redefine the capabilities of what AI can create and understand, ultimately accelerating technological progress and expanding the frontier of human-machine collaboration.
Alexia strongly believes in the power of efficiency and elegance in AI design—that smaller, smarter models can outperform bulky, inefficient counterparts. She values transparency, open sharing of knowledge (as evidenced by sharing papers and code), and continuous exploration of novel methods to tackle complex reasoning tasks. Innovation, rigor, and collaboration underpin her approach to research and impact.
Exceptional technical expertise in AI research, combined with a visionary approach to designing compact, high-performing models. Highly skilled communicator who effectively shares complex ideas through code, blogs, and concise tweets that captivate and educate a diverse audience.
A deep focus on technical depth might make her content a bit niche or difficult for broader audiences to immediately grasp, possibly limiting widespread virality outside AI-focused circles. Also, a high volume of tweets and following could occasionally dilute engagement impact.
To grow her audience on X, Alexia should continue simplifying her explanations with relatable analogies and engaging visuals, making her groundbreaking work more accessible to enthusiasts and potential collaborators outside hardcore AI research. Engaging more with community questions and spotlighting real-world applications of her tiny models could attract a larger, cross-disciplinary following.
Fun fact: Alexia built a neural network with only 7 million parameters that beats much larger language models on advanced reasoning benchmarks—proving big things really do come in tiny packages!
{"data":{"__meta":{"device":false,"path":"/creators/jm_alexia"},"/creators/jm_alexia":{"data":{"user":{"id":"839820726777544705","name":"Alexia Jolicoeur-Martineau","description":"Senior AI Researcher at the Samsung SAIT AI Lab 🐱💻\n\nI build generative AI for images, videos, text, tabular data, weights, molecules, and video games.","followers_count":19654,"friends_count":1974,"statuses_count":9280,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1480318038427803651/i_-ZNnjh_normal.jpg","screen_name":"jm_alexia","location":"Montréal, Québec","entities":{"url":{"urls":[{"url":"https://t.co/yThXRizO5J","expanded_url":"http://ajolicoeur.wordpress.com","display_url":"ajolicoeur.wordpress.com","indices":[0,23]}]},"description":{"hashtags":[],"symbols":[],"user_mentions":[],"urls":[]}}},"details":{"type":"The Innovator","description":"Alexia Jolicoeur-Martineau is a trailblazing Senior AI Researcher passionately crafting cutting-edge generative AI models across a dazzling array of data types—from images and text to molecules and video games. With a sharp mind and a knack for defying expectations, Alexia champions ultra-efficient neural networks that challenge the status quo. This profile is a powerhouse of AI innovation, blending deep technical knowledge with a drive to push boundaries.","purpose":"To revolutionize artificial intelligence by developing groundbreaking, efficient, and versatile generative models that redefine the capabilities of what AI can create and understand, ultimately accelerating technological progress and expanding the frontier of human-machine collaboration.","beliefs":"Alexia strongly believes in the power of efficiency and elegance in AI design—that smaller, smarter models can outperform bulky, inefficient counterparts. She values transparency, open sharing of knowledge (as evidenced by sharing papers and code), and continuous exploration of novel methods to tackle complex reasoning tasks. Innovation, rigor, and collaboration underpin her approach to research and impact.","facts":"Fun fact: Alexia built a neural network with only 7 million parameters that beats much larger language models on advanced reasoning benchmarks—proving big things really do come in tiny packages!","strength":"Exceptional technical expertise in AI research, combined with a visionary approach to designing compact, high-performing models. Highly skilled communicator who effectively shares complex ideas through code, blogs, and concise tweets that captivate and educate a diverse audience.","weakness":"A deep focus on technical depth might make her content a bit niche or difficult for broader audiences to immediately grasp, possibly limiting widespread virality outside AI-focused circles. Also, a high volume of tweets and following could occasionally dilute engagement impact.","recommendation":"To grow her audience on X, Alexia should continue simplifying her explanations with relatable analogies and engaging visuals, making her groundbreaking work more accessible to enthusiasts and potential collaborators outside hardcore AI research. Engaging more with community questions and spotlighting real-world applications of her tiny models could attract a larger, cross-disciplinary following.","roast":"Alexia’s obsession with tiny neural nets shows she’s basically an AI minimalist—why build a mansion when you can just stick to an ultra-efficient shoebox stacked with layers? The real question is, does she measure success in parameter count or how many coffee cups she’s consumed during coding marathons?","win":"Achieving state-of-the-art results on complex AI benchmarks with a neural network less than 10 million parameters is a monumental breakthrough, demonstrating profound mastery and innovation that challenges the norms of AI model scaling."},"tweets":[{"bookmarked":false,"display_text_range":[0,278],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"alexiajm.github.io/2025/09/29/tin…","expanded_url":"https://alexiajm.github.io/2025/09/29/tiny_recursive_models.html","url":"https://t.co/w5ZDsHDDPE","indices":[192,215]},{"display_url":"github.com/SamsungSAILMon…","expanded_url":"https://github.com/SamsungSAILMontreal/TinyRecursiveModels","url":"https://t.co/7UgKuD9Yll","indices":[223,246]},{"display_url":"arxiv.org/abs/2510.04871","expanded_url":"https://arxiv.org/abs/2510.04871","url":"https://t.co/3m8ANhNMiw","indices":[255,278]}],"user_mentions":[]},"favorited":true,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1975560628657164426","view_count":630275,"bookmark_count":2492,"created_at":1759845332000,"favorite_count":3970,"quote_count":159,"reply_count":140,"retweet_count":661,"user_id_str":"839820726777544705","conversation_id_str":"1975560628657164426","full_text":"New paper 📜: Tiny Recursion Model (TRM) is a recursive reasoning approach with a tiny 7M parameters neural network that obtains 45% on ARC-AGI-1 and 8% on ARC-AGI-2, beating most LLMs.\n\nBlog: https://t.co/w5ZDsHDDPE\n\nCode: https://t.co/7UgKuD9Yll\n\nPaper: https://t.co/3m8ANhNMiw","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,99],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1967300931973992698","view_count":225176,"bookmark_count":768,"created_at":1757876067000,"favorite_count":2700,"quote_count":13,"reply_count":154,"retweet_count":75,"user_id_str":"839820726777544705","conversation_id_str":"1967300931973992698","full_text":"I'm working on super tiny neural networks that perform better than the big ones, it's kinda insane.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,39],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1976052730356695554","view_count":163973,"bookmark_count":205,"created_at":1759962658000,"favorite_count":1480,"quote_count":20,"reply_count":83,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1976052730356695554","full_text":"7M model coming to crush all benchmarks","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,205],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1976292337547641313","view_count":349138,"bookmark_count":432,"created_at":1760019785000,"favorite_count":1028,"quote_count":17,"reply_count":54,"retweet_count":68,"user_id_str":"839820726777544705","conversation_id_str":"1976292337547641313","full_text":"To generate 1K tokens, an LLM with 100 layers needs to go through 100*1,000 = 100,000 layers\n\nTo generate any amount of tokens, TRM do 16 reasoning steps of 2*7*3=42 depth for a total of 16*42 = 672 layers","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,67],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/GXhHtnJ7Lh","expanded_url":"https://x.com/jm_alexia/status/1956466240186126763/photo/1","id_str":"1956465777462116352","indices":[68,91],"media_key":"3_1956465777462116352","media_url_https":"https://pbs.twimg.com/media/GybDQkDWgAA-rvr.png","type":"photo","url":"https://t.co/GXhHtnJ7Lh","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":29,"y":232,"h":53,"w":53}]},"medium":{"faces":[{"x":29,"y":232,"h":53,"w":53}]},"small":{"faces":[{"x":25,"y":201,"h":45,"w":45}]},"orig":{"faces":[{"x":29,"y":232,"h":53,"w":53}]}},"sizes":{"large":{"h":783,"w":757,"resize":"fit"},"medium":{"h":783,"w":757,"resize":"fit"},"small":{"h":680,"w":657,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":783,"width":757,"focus_rects":[{"x":0,"y":120,"w":757,"h":424},{"x":0,"y":0,"w":757,"h":757},{"x":70,"y":0,"w":687,"h":783},{"x":293,"y":0,"w":392,"h":783},{"x":0,"y":0,"w":757,"h":783}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1956465777462116352"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/GXhHtnJ7Lh","expanded_url":"https://x.com/jm_alexia/status/1956466240186126763/photo/1","id_str":"1956465777462116352","indices":[68,91],"media_key":"3_1956465777462116352","media_url_https":"https://pbs.twimg.com/media/GybDQkDWgAA-rvr.png","type":"photo","url":"https://t.co/GXhHtnJ7Lh","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":29,"y":232,"h":53,"w":53}]},"medium":{"faces":[{"x":29,"y":232,"h":53,"w":53}]},"small":{"faces":[{"x":25,"y":201,"h":45,"w":45}]},"orig":{"faces":[{"x":29,"y":232,"h":53,"w":53}]}},"sizes":{"large":{"h":783,"w":757,"resize":"fit"},"medium":{"h":783,"w":757,"resize":"fit"},"small":{"h":680,"w":657,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":783,"width":757,"focus_rects":[{"x":0,"y":120,"w":757,"h":424},{"x":0,"y":0,"w":757,"h":757},{"x":70,"y":0,"w":687,"h":783},{"x":293,"y":0,"w":392,"h":783},{"x":0,"y":0,"w":757,"h":783}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1956465777462116352"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1956431617951740044","quoted_status_permalink":{"url":"https://t.co/V43wt7Dq4d","expanded":"https://twitter.com/arcprize/status/1956431617951740044","display":"x.com/arcprize/statu…"},"retweeted":false,"fact_check":null,"id":"1956466240186126763","view_count":73606,"bookmark_count":564,"created_at":1755292875000,"favorite_count":763,"quote_count":2,"reply_count":13,"retweet_count":87,"user_id_str":"839820726777544705","conversation_id_str":"1956466240186126763","full_text":"Summary of the findings for the Hierarchical Reasoning Model on ARC https://t.co/GXhHtnJ7Lh","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,299],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1965253577221587218","quoted_status_permalink":{"url":"https://t.co/spuimuv3SM","expanded":"https://twitter.com/arankomatsuzaki/status/1965253577221587218","display":"x.com/arankomatsuzak…"},"retweeted":false,"fact_check":null,"id":"1967214949861904764","view_count":76666,"bookmark_count":338,"created_at":1757855567000,"favorite_count":722,"quote_count":5,"reply_count":19,"retweet_count":87,"user_id_str":"839820726777544705","conversation_id_str":"1967214949861904764","full_text":"Researcher in 2026\n>Use LLMs to generate 20 papers\n>Spend 2 weeks clicking approve/reject\n>Bill: 20k$ in API credits\n>Spend 2 weeks verifying them, 1 is good\n>Publish paper\n>GitHub issues: Data leakage, bugs; ignore\n>post on X: Human coding is obsolete! AI invented new science🤯","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,210],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1151388211752636418","name":"Konstantin Schürholt","screen_name":"k_schuerholt","indices":[48,61]},{"id_str":"1773935160192647168","name":"ARC Prize","screen_name":"arcprize","indices":[78,87]},{"id_str":"1773935160192647168","name":"ARC Prize","screen_name":"arcprize","indices":[198,207]}]},"favorited":false,"lang":"en","quoted_status_id_str":"1978872651180577060","quoted_status_permalink":{"url":"https://t.co/FPjG7oVLVy","expanded":"https://twitter.com/arcprize/status/1978872651180577060","display":"x.com/arcprize/statu…"},"retweeted":false,"fact_check":null,"id":"1978882845801128299","view_count":55253,"bookmark_count":77,"created_at":1760637410000,"favorite_count":491,"quote_count":6,"reply_count":16,"retweet_count":45,"user_id_str":"839820726777544705","conversation_id_str":"1978882845801128299","full_text":"I wanna grab your attention one moment to thank @k_schuerholt and the team at @arcprize for allowing me to participate in this historic competition normally only reserved for the giants.\n\nThank you @arcprize !!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,67],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1979066769814802886","quoted_status_permalink":{"url":"https://t.co/aIz8fffp4S","expanded":"https://twitter.com/tokenbender/status/1979066769814802886","display":"x.com/tokenbender/st…"},"retweeted":false,"fact_check":null,"id":"1979581330162876753","view_count":73491,"bookmark_count":187,"created_at":1760803942000,"favorite_count":423,"quote_count":4,"reply_count":19,"retweet_count":20,"user_id_str":"839820726777544705","conversation_id_str":"1979581330162876753","full_text":"4.2 mil USD to solve 6 extra math problems. Seems highly efficient!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,262],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1871308418390364209","quoted_status_permalink":{"url":"https://t.co/BrB4wZNPHM","expanded":"https://twitter.com/fchollet/status/1871308418390364209","display":"x.com/fchollet/statu…"},"retweeted":false,"fact_check":null,"id":"1871313773333741964","view_count":56604,"bookmark_count":183,"created_at":1734990945000,"favorite_count":415,"quote_count":6,"reply_count":12,"retweet_count":18,"user_id_str":"839820726777544705","conversation_id_str":"1871313773333741964","full_text":"2 days ago, people were trashing me for saying that \"Inference-time scaling is effectively genetic algorithms with LLMs.\"\n\n#1 on ARC-AGI-Pub today: \"I was inspired by genetic algorithms and have been referring to this approach as Evolutionary Test-time Compute.\"","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,76],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"huggingface.co/papers/date/20…","expanded_url":"https://huggingface.co/papers/date/2025-10-08","url":"https://t.co/vQOyq5QOrX","indices":[53,76]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1975982176744374310","view_count":132072,"bookmark_count":111,"created_at":1759945837000,"favorite_count":404,"quote_count":5,"reply_count":23,"retweet_count":35,"user_id_str":"839820726777544705","conversation_id_str":"1975982176744374310","full_text":"TRM is now the #1 trending paper on the Daily Papers https://t.co/vQOyq5QOrX","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/orxFJML3Gl","expanded_url":"https://x.com/jm_alexia/status/1971563912115200328/photo/1","id_str":"1971563578877526017","indices":[26,49],"media_key":"3_1971563578877526017","media_url_https":"https://pbs.twimg.com/media/G1xmoRMXwAELuhq.jpg","type":"photo","url":"https://t.co/orxFJML3Gl","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1436,"y":389,"h":73,"w":73},{"x":1038,"y":389,"h":76,"w":76}]},"medium":{"faces":[{"x":1011,"y":273,"h":51,"w":51},{"x":730,"y":273,"h":53,"w":53}]},"small":{"faces":[{"x":573,"y":155,"h":29,"w":29},{"x":414,"y":155,"h":30,"w":30}]},"orig":{"faces":[{"x":1436,"y":389,"h":73,"w":73},{"x":1038,"y":389,"h":76,"w":76}]}},"sizes":{"large":{"h":639,"w":1704,"resize":"fit"},"medium":{"h":450,"w":1200,"resize":"fit"},"small":{"h":255,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":639,"width":1704,"focus_rects":[{"x":0,"y":0,"w":1141,"h":639},{"x":0,"y":0,"w":639,"h":639},{"x":0,"y":0,"w":561,"h":639},{"x":53,"y":0,"w":320,"h":639},{"x":0,"y":0,"w":1704,"h":639}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1971563578877526017"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/orxFJML3Gl","expanded_url":"https://x.com/jm_alexia/status/1971563912115200328/photo/1","id_str":"1971563578877526017","indices":[26,49],"media_key":"3_1971563578877526017","media_url_https":"https://pbs.twimg.com/media/G1xmoRMXwAELuhq.jpg","type":"photo","url":"https://t.co/orxFJML3Gl","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1436,"y":389,"h":73,"w":73},{"x":1038,"y":389,"h":76,"w":76}]},"medium":{"faces":[{"x":1011,"y":273,"h":51,"w":51},{"x":730,"y":273,"h":53,"w":53}]},"small":{"faces":[{"x":573,"y":155,"h":29,"w":29},{"x":414,"y":155,"h":30,"w":30}]},"orig":{"faces":[{"x":1436,"y":389,"h":73,"w":73},{"x":1038,"y":389,"h":76,"w":76}]}},"sizes":{"large":{"h":639,"w":1704,"resize":"fit"},"medium":{"h":450,"w":1200,"resize":"fit"},"small":{"h":255,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":639,"width":1704,"focus_rects":[{"x":0,"y":0,"w":1141,"h":639},{"x":0,"y":0,"w":639,"h":639},{"x":0,"y":0,"w":561,"h":639},{"x":53,"y":0,"w":320,"h":639},{"x":0,"y":0,"w":1704,"h":639}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1971563578877526017"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1963635422698856705","quoted_status_permalink":{"url":"https://t.co/COPRigFACA","expanded":"https://twitter.com/GoogleDeepMind/status/1963635422698856705","display":"x.com/GoogleDeepMind…"},"retweeted":false,"fact_check":null,"id":"1971563912115200328","view_count":58278,"bookmark_count":194,"created_at":1758892441000,"favorite_count":373,"quote_count":4,"reply_count":11,"retweet_count":25,"user_id_str":"839820726777544705","conversation_id_str":"1971563912115200328","full_text":"BERT wins, tiny is better https://t.co/orxFJML3Gl","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,282],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1880322648980484178","quoted_status_permalink":{"url":"https://t.co/n0IjJY9Uad","expanded":"https://twitter.com/mgostIH/status/1880322648980484178","display":"x.com/mgostIH/status…"},"retweeted":false,"fact_check":null,"id":"1880654503080530019","view_count":41429,"bookmark_count":257,"created_at":1737217949000,"favorite_count":369,"quote_count":2,"reply_count":7,"retweet_count":29,"user_id_str":"839820726777544705","conversation_id_str":"1880654503080530019","full_text":"This is my favorite paper of 2025 so far. \n\n\"Hyperfitting\": When a language model overfits (train loss -> 0, eval loss increases over time), greedy (top-1) decoding leads to high-quality and diverse (non-copied) generated samples. \n\nThis is so counterintuitive, it feels magical.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":21709,"bookmark_count":231,"created_at":1762264315000,"favorite_count":362,"quote_count":3,"reply_count":8,"retweet_count":48,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}],"ctweets":[{"bookmarked":false,"display_text_range":[0,99],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1967300931973992698","view_count":225176,"bookmark_count":768,"created_at":1757876067000,"favorite_count":2700,"quote_count":13,"reply_count":154,"retweet_count":75,"user_id_str":"839820726777544705","conversation_id_str":"1967300931973992698","full_text":"I'm working on super tiny neural networks that perform better than the big ones, it's kinda insane.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,278],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"alexiajm.github.io/2025/09/29/tin…","expanded_url":"https://alexiajm.github.io/2025/09/29/tiny_recursive_models.html","url":"https://t.co/w5ZDsHDDPE","indices":[192,215]},{"display_url":"github.com/SamsungSAILMon…","expanded_url":"https://github.com/SamsungSAILMontreal/TinyRecursiveModels","url":"https://t.co/7UgKuD9Yll","indices":[223,246]},{"display_url":"arxiv.org/abs/2510.04871","expanded_url":"https://arxiv.org/abs/2510.04871","url":"https://t.co/3m8ANhNMiw","indices":[255,278]}],"user_mentions":[]},"favorited":true,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1975560628657164426","view_count":630275,"bookmark_count":2492,"created_at":1759845332000,"favorite_count":3970,"quote_count":159,"reply_count":140,"retweet_count":661,"user_id_str":"839820726777544705","conversation_id_str":"1975560628657164426","full_text":"New paper 📜: Tiny Recursion Model (TRM) is a recursive reasoning approach with a tiny 7M parameters neural network that obtains 45% on ARC-AGI-1 and 8% on ARC-AGI-2, beating most LLMs.\n\nBlog: https://t.co/w5ZDsHDDPE\n\nCode: https://t.co/7UgKuD9Yll\n\nPaper: https://t.co/3m8ANhNMiw","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,39],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1976052730356695554","view_count":163973,"bookmark_count":205,"created_at":1759962658000,"favorite_count":1480,"quote_count":20,"reply_count":83,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1976052730356695554","full_text":"7M model coming to crush all benchmarks","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,205],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1976292337547641313","view_count":349138,"bookmark_count":432,"created_at":1760019785000,"favorite_count":1028,"quote_count":17,"reply_count":54,"retweet_count":68,"user_id_str":"839820726777544705","conversation_id_str":"1976292337547641313","full_text":"To generate 1K tokens, an LLM with 100 layers needs to go through 100*1,000 = 100,000 layers\n\nTo generate any amount of tokens, TRM do 16 reasoning steps of 2*7*3=42 depth for a total of 16*42 = 672 layers","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,76],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"huggingface.co/papers/date/20…","expanded_url":"https://huggingface.co/papers/date/2025-10-08","url":"https://t.co/vQOyq5QOrX","indices":[53,76]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1975982176744374310","view_count":132072,"bookmark_count":111,"created_at":1759945837000,"favorite_count":404,"quote_count":5,"reply_count":23,"retweet_count":35,"user_id_str":"839820726777544705","conversation_id_str":"1975982176744374310","full_text":"TRM is now the #1 trending paper on the Daily Papers https://t.co/vQOyq5QOrX","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,252],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":true,"lang":"en","retweeted":true,"fact_check":null,"id":"1814309242129535386","view_count":71652,"bookmark_count":253,"created_at":1721400006000,"favorite_count":345,"quote_count":3,"reply_count":22,"retweet_count":23,"user_id_str":"839820726777544705","conversation_id_str":"1814309242129535386","full_text":"Instead of LLMs, why not try molecule generation? \n\nDatasets are much smaller with low sequence length, which means that Transformers can be trained in reasonable on 1 or few GPUs. So if you are GPU poor, try it out. It also has more real-world impact.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,68],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1914006959411409158","view_count":22478,"bookmark_count":35,"created_at":1745169794000,"favorite_count":133,"quote_count":3,"reply_count":20,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1914006959411409158","full_text":"🤯I figured out an alternative to Transformers, it might even work!!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,299],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1965253577221587218","quoted_status_permalink":{"url":"https://t.co/spuimuv3SM","expanded":"https://twitter.com/arankomatsuzaki/status/1965253577221587218","display":"x.com/arankomatsuzak…"},"retweeted":false,"fact_check":null,"id":"1967214949861904764","view_count":76666,"bookmark_count":338,"created_at":1757855567000,"favorite_count":722,"quote_count":5,"reply_count":19,"retweet_count":87,"user_id_str":"839820726777544705","conversation_id_str":"1967214949861904764","full_text":"Researcher in 2026\n>Use LLMs to generate 20 papers\n>Spend 2 weeks clicking approve/reject\n>Bill: 20k$ in API credits\n>Spend 2 weeks verifying them, 1 is good\n>Publish paper\n>GitHub issues: Data leakage, bugs; ignore\n>post on X: Human coding is obsolete! AI invented new science🤯","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,67],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1979066769814802886","quoted_status_permalink":{"url":"https://t.co/aIz8fffp4S","expanded":"https://twitter.com/tokenbender/status/1979066769814802886","display":"x.com/tokenbender/st…"},"retweeted":false,"fact_check":null,"id":"1979581330162876753","view_count":73491,"bookmark_count":187,"created_at":1760803942000,"favorite_count":423,"quote_count":4,"reply_count":19,"retweet_count":20,"user_id_str":"839820726777544705","conversation_id_str":"1979581330162876753","full_text":"4.2 mil USD to solve 6 extra math problems. Seems highly efficient!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,210],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1151388211752636418","name":"Konstantin Schürholt","screen_name":"k_schuerholt","indices":[48,61]},{"id_str":"1773935160192647168","name":"ARC Prize","screen_name":"arcprize","indices":[78,87]},{"id_str":"1773935160192647168","name":"ARC Prize","screen_name":"arcprize","indices":[198,207]}]},"favorited":false,"lang":"en","quoted_status_id_str":"1978872651180577060","quoted_status_permalink":{"url":"https://t.co/FPjG7oVLVy","expanded":"https://twitter.com/arcprize/status/1978872651180577060","display":"x.com/arcprize/statu…"},"retweeted":false,"fact_check":null,"id":"1978882845801128299","view_count":55253,"bookmark_count":77,"created_at":1760637410000,"favorite_count":491,"quote_count":6,"reply_count":16,"retweet_count":45,"user_id_str":"839820726777544705","conversation_id_str":"1978882845801128299","full_text":"I wanna grab your attention one moment to thank @k_schuerholt and the team at @arcprize for allowing me to participate in this historic competition normally only reserved for the giants.\n\nThank you @arcprize !!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"truncated":false,"display_text_range":[0,228],"entities":{"hashtags":[],"symbols":[],"user_mentions":[],"urls":[]},"source":"<a href=\"https://mobile.twitter.com\" rel=\"nofollow\">Twitter Web App</a>","in_reply_to_status_id":null,"in_reply_to_user_id":null,"in_reply_to_screen_name":null,"user_id":839820726777544700,"geo":null,"coordinates":null,"place":null,"contributors":null,"conversation_id":1856420861336486400,"conversation_muted":false,"favorited":false,"retweeted":false,"lang":"en","ext":{"superFollowMetadata":{"r":{"ok":{}},"ttl":-1}},"fact_check":null,"id":"1856420861336486392","view_count":50791,"bookmark_count":null,"created_at":1731440198000,"favorite_count":285,"quote_count":3,"reply_count":14,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1856420861336486392","full_text":"ICLR reviews are out and they are horrendous. \n\nAlso, papers have 5-7 reviews! This is unreasonable. Researchers have other things to do than answer 7 reviewers. This is more than double the work compared to the usual 3 reviews.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,277],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1816862928936501374","view_count":37908,"bookmark_count":117,"created_at":1722008852000,"favorite_count":186,"quote_count":0,"reply_count":13,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1816862928936501374","full_text":"Does anyone know why autoregressive models like LLMs never use Exponential Moving Average (EMA)? \n\nDiffusion models and GANs get a big boost in performance from it.\n\nIs it because the probabilities of the next token are relatively stable through time, so there is no perf gain?","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}],"activities":{"nreplies":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":0,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":4,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":19,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":19,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":0,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":14,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":9,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":4,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":10,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":0,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":9,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":0,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":67,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":15,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}],"nbookmarks":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":1,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":12,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":348,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":44,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":68,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":353,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":96,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":31,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":377,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":1,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":96,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":0,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":140,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":432,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}],"nretweets":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":0,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":4,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":48,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":18,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":9,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":67,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":8,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":3,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":69,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":0,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":11,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":0,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":22,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":46,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}],"nlikes":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":3,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":53,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":580,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":249,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":136,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":575,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":273,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":125,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":576,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":8,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":308,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":1,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":592,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":618,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}],"nviews":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":404,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":14223,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":78556,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":11259,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":29828,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":51133,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":63189,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":18886,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":37770,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":217,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":43956,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":48,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":80853,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":81809,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}]},"interactions":{"users":[],"period":14,"start":1762385183280,"end":1763594783280},"interactions_updated":1763594783345,"created":1763362814260,"updated":1763594783345,"type":"the innovator","hits":2},"people":[{"user":{"id":"1912602722","name":"Sophie","description":"🌸 Web3女战士｜#Binance Alpha猎人｜超级纸手｜BRC-20、NFT及DeFi挖矿｜🐱撸猫控，游戏和链上机会的热情追逐者｜专注币圈一级市场及早期项目研究｜#Bitcoin #ETH #Web3 #DAO｜欢迎私信合作对接💌｜","followers_count":107010,"friends_count":3859,"statuses_count":16004,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1853830480157986816/xWbEHNnd_normal.jpg","screen_name":"hooe9212","location":"Hong Kong","entities":{"description":{"urls":[]}}},"details":{"type":"The Innovator","description":"Sophie is a pioneering Web3 warrior deeply embedded in the crypto trenches, passionately hunting alpha opportunities across BRC-20, NFTs, and DeFi mining. She combines sharp early-stage market insights with a love for gaming and blockchain tech, always on the hunt for the next big breakout in the decentralized economy. Her tweets decode complex crypto ecosystems with clarity and real-world passion.","purpose":"Sophie's life purpose is to bridge the gap between emerging blockchain technologies and everyday users by championing innovation and driving meaningful adoption in the crypto space. She seeks to empower the community to leverage early opportunities and create value through cutting-edge decentralized finance and Web3 projects.","beliefs":"She strongly believes in transparency, genuine value creation, and the transformative power of blockchain to redistribute wealth and democratize access. Sophie holds that sustainable growth in crypto stems not from hype but from real-world impact, measurable data, and user-centric innovation.","facts":"Fun fact: She’s a self-confessed '超级纸手' (super weak hands) crypto enthusiast, yet remains incredibly prolific, tweeting over 16,000 times to share insights on projects that prevent ‘sheep farming’ and cultivate real alpha!","strength":"Sophie’s key strengths lie in her relentless research of early-stage market projects, thoughtful breakdowns of complex DeFi protocols, and active engagement with high-value community partnerships. Her ability to synthesize industry narratives with quantifiable on-chain data makes her a trusted source in her niche.","weakness":"Her '超级纸手' nature might lead to missing out on long-term holdings, and her high tweet volume could potentially overwhelm audiences or dilute her most insightful content among the noise.","recommendation":"To grow her audience on X, Sophie should strategically leverage thread storytelling to deepen her educational content while curating her best analyses into highlight tweets. Engaging more with emerging influencers and hosting Twitter Spaces focused on early-stage Web3 opportunities can amplify her thought leadership.","roast":"Sophie’s like a crypto grasshopper – hopping from one shiny token to the next so fast, even her cats need a blockchain to keep track of her attention span. At this rate, her portfolio’s probably got more splits than a yoga class.","win":"Sophie’s biggest win is building a highly engaged and authentic Web3 community by being one of the early voices to spotlight sustainable DeFi projects and real-world crypto integrations, earning trust and collaborations with leading ecosystem players like Binance Alpha and major protocols."},"created":1763364880831,"type":"the innovator","id":"hooe9212"},{"user":{"id":"14830733","name":"Simon Hamp","description":"@nativephp","followers_count":6331,"friends_count":426,"statuses_count":23104,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1963678298539450368/UCDiig_S_normal.jpg","screen_name":"simonhamp","location":"Las Palmas de Gran Canaria","entities":{"description":{"urls":[]},"url":{"urls":[{"display_url":"pinkary.com/@simonhamp","expanded_url":"https://pinkary.com/@simonhamp","url":"https://t.co/Bi8uZMamaR","indices":[0,23]}]}}},"details":{"type":"The Innovator","description":"Simon Hamp, also known as @nativephp, is a fearless tech pioneer shaking up the developer world with his groundbreaking announcements and unapologetic opinions. With over 23,000 tweets, he's a passionate advocate for PHP’s evolving capabilities, especially in mobile app development. Simon isn't just breaking molds—he’s setting the stage for new tech standards.","purpose":"Simon’s life purpose is to redefine what’s possible with PHP, challenging conventional tech boundaries and empowering developers to embrace innovative tools and platforms. He strives to position PHP as a future-proof language that commands respect beyond traditional web applications.","beliefs":"He believes in disruption, transparency, and pushing technological limits. Simon values authenticity and community-driven growth in the developer ecosystem, empowering tech enthusiasts to innovate boldly and question norms without fear.","facts":"Fun fact: Simon announced on stage at LaraconEU the very first iPhone app built using LaravelPHP and LaravelLivewire, marking a major milestone where PHP earned Apple’s approval for mobile app development.","strength":"Relentless innovator with a deep knowledge of PHP and tech ecosystems, exceptional at rallying developer communities and generating significant buzz around new tech breakthroughs.","weakness":"Sometimes his bluntness and prolific tweeting can overwhelm followers or create friction, as not every tech enthusiast may be ready for such radical shifts or critical takes.","roast":"Simon tweets so much and so fast, it’s like PHP had an espresso shot—if only debugging was that quick! At this rate, he might just tweet the next programming language into existence before anyone else even finishes their coffee.","win":"A standout achievement is his live announcement at LaraconEU revealing the first-ever iPhone app built using PHP technologies, significantly elevating PHP’s credibility in mobile development circles.","recommendation":"To grow his audience on X, Simon should blend his technical depth with even more engaging storytelling and community interaction—think live coding sessions, AMA threads, and strategic collaborations with influencers in mobile and web tech spheres to amplify his visionary message."},"created":1763364767860,"type":"the innovator","id":"simonhamp"},{"user":{"id":"1396100302206603267","name":"Oxxyy","description":"CPA | DeFi | Be Curious\n\nUse: \"Oxxyy\" on Solstice","followers_count":6737,"friends_count":5020,"statuses_count":46211,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1943066795197202432/XXwhGtSk_normal.jpg","screen_name":"Oxxyy13","location":"","entities":{"description":{"urls":[]},"url":{"urls":[{"display_url":"prjx.com/@Oxxyy","expanded_url":"https://www.prjx.com/@Oxxyy","url":"https://t.co/rFEOHKOfdn","indices":[0,23]}]}}},"details":{"type":"The Innovator","description":"Oxxyy is a dynamic DeFi enthusiast and CPA who thrives on exploring new financial technologies and maximizing digital asset strategies. Their high tweet volume and strategic insights into Web3 demonstrate a deep commitment to empowering their community with actionable knowledge. Always curious and forward-thinking, Oxxyy bridges the gap between traditional finance expertise and cutting-edge crypto innovation with a uniquely engaging voice.","purpose":"To demystify decentralized finance and lead the Web3 community toward smarter, more profitable financial decisions while inspiring curiosity and continuous learning.","beliefs":"Oxxyy believes in the transformative power of decentralization, continual curiosity, and the importance of combining professional rigor with innovative thinking in the evolving digital economy.","facts":"Oxxyy has tweeted over 46,000 times, showing relentless engagement and passion for Web3 topics, and recently accepted a Director position at 0n1Force, marking a major career milestone in the space.","strength":"Oxxyy's strengths lie in their deep technical knowledge, ability to break down complex DeFi concepts into practical strategies, and a high-energy presence that keeps their audience both informed and entertained.","weakness":"With a massive tweet output, Oxxyy risks overwhelming followers with information or appearing scattered, which may dilute the impact of their most valuable insights.","roast":"Oxxyy tweets so much, their keyboard probably files for overtime – if enthusiasm translated to follower count, they'd be the richest wallet on the blockchain by now!","win":"Securing a leadership role as Director at 0n1Force after consistently proving expertise and influence within the Web3 community.","recommendation":"To grow their audience on X, Oxxyy should consider curating some of their vast content into themed mini-series or digestible threads highlighting key strategies, allowing new followers to grasp core insights without the need to sift through thousands of tweets. Plus, interactive AMAs or polls on DeFi topics could boost engagement and position Oxxyy as the go-to Web3 innovator."},"created":1763364346574,"type":"the innovator","id":"oxxyy13"},{"user":{"id":"7005032","name":"Kath Korevec","description":"Director of Product at Google Labs. Code AI. Dive in ➡ @googlelabs, @stitchbygoogle, and @julesagent Previously @vercel, @github and @heroku","followers_count":18975,"friends_count":592,"statuses_count":8692,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1899487081233252352/ZORhD9Nt_normal.jpg","screen_name":"simpsoka","location":"","entities":{"description":{"urls":[]},"url":{"urls":[{"display_url":"kathy.pm","expanded_url":"https://kathy.pm/","url":"https://t.co/smXbQDlXIc","indices":[0,23]}]}}},"details":{"type":"The Innovator","description":"Kath Korevec is a forward-thinking product director at Google Labs, blending cutting-edge AI coding with an adventurous spirit that fuels tech innovation. With a passion for creating and sharing, Kath effortlessly marries technical mastery with engaging storytelling. Their tweets reveal someone who takes pride in impactful work and thrives on pushing the boundaries of technology while staying grounded in integrity.","purpose":"Kath’s life purpose is to drive technological progress by creating pioneering products that not only advance the future but also uphold ethical standards and inclusivity. They aim to inspire others in tech to innovate boldly while maintaining community trust and transparency.","beliefs":"Kath believes that innovation and integrity go hand in hand, valuing transparency, collaboration, and meaningful impact over mere hype. They hold that technology should enhance human potential and be accessible and responsible.","facts":"Fun fact: Kath built a giant JavaScript-powered flipdisc display for their office, showing a playful yet expert approach to blending hardware and software creativity.","strength":"Kath’s strengths lie in visionary product leadership, technical expertise in AI, and a genuine enthusiasm for sharing knowledge with an engaged community. Their ability to cite real-world projects and insightful commentary highlights a rare balance of depth and approachability.","weakness":"Their high tweet volume (8,692 tweets) and broad following count (following 592) might dilute the focus of their messaging, occasionally making it harder to stand out in a saturated tech space or create highly targeted engagement.","recommendation":"To grow their audience on X, Kath should leverage more focused storytelling around signature projects like Google Labs and their flipdisc display, use threads to dive deeper, and engage more with niche tech communities and influencers to amplify reach and foster stronger connection.","roast":"Kath tweets so much you could swear they’re single-handedly keeping the character count economy afloat — a relentless coder who moonlights as an accidental Twitter novelist!","win":"Kath’s biggest win is leading innovative AI projects at Google Labs while successfully sharing complex tech in relatable ways that resonate widely, making cutting-edge research accessible and exciting."},"created":1763364112509,"type":"the innovator","id":"simpsoka"},{"user":{"id":"10065202","name":"Sixia \"Leask\" Huang","description":"break the wall or bring the war","followers_count":20423,"friends_count":1759,"statuses_count":104302,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1905074285066002432/Si2NthqU_normal.jpg","screen_name":"LeaskH","location":"Ottawa, Ontario","entities":{"description":{"urls":[]},"url":{"urls":[{"display_url":"leaskh.com","expanded_url":"https://leaskh.com","url":"https://t.co/JR8qOnNtYn","indices":[0,23]}]}}},"details":{"type":"The Innovator","description":"Sixia 'Leask' Huang is a tech-savvy, boundary-breaking innovator who thrives on pushing limits and exploring cutting-edge solutions. With a tireless tweeting frequency and keen insights into digital ecosystems, Leask blends technical savvy with a rebellious spirit. Their feed is a captivating mix of practical tech talk, societal commentary, and personal moments that break norms and invite engagement.","purpose":"Leask’s life purpose revolves around dismantling barriers—both in technology and society—and pioneering new paths for others to follow. They aim to disrupt complacency and empower their audience with fresh ideas and practical hacks that challenge the status quo.","beliefs":"Leask values transparency, resilience, and innovation, believing that meaningful progress requires breaking through traditional constraints and embracing risk. They mistrust outdated systems and champion the use of technology as a tool for liberation and empowerment.","facts":"Fun fact: Despite tweeting over 104,000 times, Leask’s engagement numbers suggest a highly dedicated niche audience that deeply values their content, reflecting a persistent and influential presence rather than mass popularity.","strength":"Relentless content creator with a unique voice, deep technical knowledge, and the ability to blend serious digital insights with relatable, personal storytelling.","weakness":"Sometimes the sheer volume and niche technical detail of their tweets create a barrier to broader audience engagement, potentially limiting reach to only the most tech-literate followers.","recommendation":"To grow on X, Leask should leverage more storytelling and simplified tech insights that invite curiosity and make their expertise accessible. Engaging in conversations with influencers and connectors can amplify their reach, while thematic threads breaking down complex ideas could boost shareability.","roast":"Leask’s tweet count is so high, they’ve probably tweeted more in the last year than some people have blinked—at this rate, they might just break Twitter’s servers before breaking new ground.","win":"Consistently maintaining a fiercely original and engaged voice with over 104,000 tweets—an astounding achievement in digital persistence and innovation."},"created":1763364053782,"type":"the innovator","id":"leaskh"},{"user":{"id":"1925569963009744897","name":"Rialo","description":"Rethink. Rebuild. Rialo.\nThe only network you need.\nBacked by @PanteraCapital","followers_count":79799,"friends_count":9,"statuses_count":620,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1950265537784926208/qbjSWMDP_normal.jpg","screen_name":"RialoHQ","location":"","entities":{"description":{"urls":[]},"url":{"urls":[{"display_url":"rialo.io","expanded_url":"http://rialo.io","url":"https://t.co/grDP3pizFo","indices":[0,23]}]}}},"details":{"type":"The Innovator","description":"Rialo is a bold blockchain trailblazer dedicated to rethinking and rebuilding the future of real-world networks. With strong backing from top venture capitalists and a focus on seamless, high-speed technology, they are paving the way for next-gen digital infrastructure. Their vibrant community buzz reflects a mission to unite fragmented ideas into a powerful ecosystem.","purpose":"To revolutionize blockchain technology by creating a unified, ultra-fast network that bridges traditional finance with cutting-edge decentralization, empowering builders to take bigger swings and making sophisticated tools accessible to everyone.","beliefs":"Rialo values innovation, collaboration, and speed, believing that the future belongs to those who break down silos and rethink existing systems. They hold a strong conviction that a seamless developer experience and real-time data are keys to unlocking blockchain's full potential.","facts":"Fun fact: Despite being backed by high-profile investors like Pantera Capital and launching groundbreaking real-time data solutions, Rialo keeps a lean following, proving it’s more about quality building than follower count.","strength":"Rialo's strengths lie in its visionary leadership and strong financial backing, enabling bold experimentation and rapid development. Their ability to attract top blockchain pioneers and launch highly technical products with real world impact sets them apart.","weakness":"Rialo's niche, technically complex focus might limit mass appeal, and their current small following could slow network effects essential to blockchain ecosystems. Communication can sometimes lean heavily on jargon, potentially alienating newcomers.","recommendation":"To grow their audience on X, Rialo should leverage storytelling to humanize their tech, posting behind-the-scenes glimpses, user success stories, and simplifying complex concepts. Engaging more actively in industry conversations and hosting Q&A threads would turn their quiet following into a buzzing community.","roast":"For a project promising to rethink and rebuild the blockchain world, Rialo's social game feels like it’s still stuck in beta—maybe their network is ultra-fast, but their tweet charisma definitely needs an upgrade to break through the noise!","win":"Rialo's biggest win is securing a $20M initial funding round led by Pantera Capital and assembling a powerhouse of blockchain pioneers to back and build the first Real World blockchain network, setting a solid foundation for future successes."},"created":1763363696343,"type":"the innovator","id":"rialohq"},{"user":{"id":"1957159024635359232","name":"Thaha Wahid","description":"Top Rated @upwork | https://t.co/7KZTzkSzvn\nAutomation Engineer for real businesses, not unicorns\nCRM | GoogleScripts | Airtable | Make | N8N | Zoho","followers_count":65,"friends_count":118,"statuses_count":260,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1975155988752666624/zLW-Krlj_normal.jpg","screen_name":"thahawahid","location":"United Kingdom","entities":{"url":{"urls":[{"url":"https://t.co/URLYKEU1wl","expanded_url":"http://nisabms.com","display_url":"nisabms.com","indices":[0,23]}]},"description":{"urls":[{"url":"https://t.co/7KZTzkSzvn","expanded_url":"https://www.nisabms.com/uw","display_url":"nisabms.com/uw","indices":[20,43]}]}}},"details":{"type":"The Innovator","description":"Thaha Wahid is a hands-on Automation Engineer who thrives on creating practical solutions for real businesses using tools like Airtable, Google Scripts, and Make. He's a builder at heart, constantly pushing projects like NisaNative—a B2B app converting Airtable into a native mobile experience. With a clear focus on progress over perfection, Thaha embodies the spirit of innovation and relentless distribution.","purpose":"To empower businesses by transforming complex automation and CRM challenges into streamlined, user-friendly tools that drive efficiency and real-world impact.","beliefs":"Thaha believes that launching a functional solution trumps endless tweaking of perfect drafts and that practical, scalable automation can revolutionize how businesses operate. He values continuous progress, resourcefulness, and making technology accessible for everyday users rather than chasing unicorn status.","facts":"Fun fact: Thaha isn’t just building automation workflows; he’s creating entire apps that let teams and clients interact directly with Airtable data through native mobile experiences—no unicorns required!","strength":"His biggest strength lies in the ability to rapidly prototype and launch working products, seamlessly integrating multiple tech platforms to solve real business problems with practical automation.","weakness":"Sometimes the relentless push for distribution and rapid iteration might overshadow polishing the user experience or marketing finesse, which could limit the early audience impact.","roast":"Thaha is that rare developer who launches apps so quickly, you half expect bugs to RSVP themselves — but hey, who needs perfection when you’ve got caffeine and enough tokens left to keep the bots running?","win":"Successfully building and progressing NisaNative, a complex B2B app that effectively turns Airtable bases into authentic native mobile apps, showcasing his technical prowess and innovative mindset.","recommendation":"To grow his audience on X, Thaha should share more behind-the-scenes content about his automation projects and challenges, combined with actionable tips, quick hacks, and tutorials. Engaging with niche communities around Airtable and no-code tools with regular, accessible threads can also establish him as a go-to innovator in this space."},"created":1763363043782,"type":"the innovator","id":"thahawahid"},{"user":{"id":"1053608268277309440","name":"Taha Tesser","description":"Kotlin Multiplatform Engineer at DoorDash/Wolt • Coffee Enthusiast • Former Design System Eng / Flutter Contributor #SaunaTeam","followers_count":6347,"friends_count":450,"statuses_count":12008,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1548238615771111424/ayLq1K_F_normal.jpg","screen_name":"TahaTesser","location":"Tallinn, Estonia","entities":{"description":{"urls":[]},"url":{"urls":[{"display_url":"designdino.dev","expanded_url":"http://designdino.dev","url":"https://t.co/k0kDAjKSW9","indices":[0,23]}]}}},"details":{"type":"The Innovator","description":"Taha Tesser is a passionate Kotlin Multiplatform Engineer who thrives at the intersection of engineering excellence and design innovation. With a rich mix of experience spanning design systems and Flutter contributions, Taha continuously pushes the boundaries of mobile and cross-platform development. Always energized by coffee and the community, he's a dynamic voice shaping modern developer tools and frameworks.","purpose":"To pioneer seamless, scalable, and well-designed cross-platform solutions that empower developers worldwide while bridging gaps between various development ecosystems.","beliefs":"Taha believes in the power of open collaboration, the importance of well-crafted design systems, and the future-forward promise of multiplatform development. He values continuous learning, community engagement, and thoughtful innovation over quick fixes.","facts":"Fun fact: Taha doesn’t just code — he’s also a former Design System Engineer and a dedicated member of the #SaunaTeam, proving that great ideas brew as well as great coffee.","strength":"Exceptional ability to integrate design sensibility into technical platforms, enthusiastic early adopter of emerging tools, and a prolific content creator who consistently engages the developer community.","weakness":"Sometimes so deeply immersed in cutting-edge tech and tooling that occasional deeper personal brand storytelling or broader industry conversations are overlooked.","roast":"Taha’s tweet storm count suggests he might single-handedly keep the Twitter servers chugging — if only his coffee intake powered his follower growth the same way it fuels his code commits!","win":"Landing a key role as Design System Engineer at Wolt, showcasing his ability to lead and innovate at the crossroads of design and development in a fast-paced, real-world environment.","recommendation":"To expand his audience on X, Taha should blend his technical insights with more personal storytelling and developer journey highlights, inviting community interactions through Q&A threads or AMAs. Leveraging trending hashtags and engaging with other influencers in Kotlin and Flutter spaces can also boost visibility."},"created":1763362415621,"type":"the innovator","id":"tahatesser"},{"user":{"id":"832594482","name":"Arthur","description":"ai @Distributed____ | https://t.co/CCt20IwXL7 prev: @Georgetown @GoldmanSachs @BrainchainAI","followers_count":2176,"friends_count":6937,"statuses_count":10068,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1972508446810345472/Nxn0C547_normal.jpg","screen_name":"arthurcolle","location":"distributed.systems","entities":{"description":{"urls":[{"display_url":"idsc2025.substack.com","expanded_url":"https://idsc2025.substack.com/","url":"https://t.co/CCt20IwXL7","indices":[22,45]}]},"url":{"urls":[{"display_url":"arthurcolle.org","expanded_url":"http://arthurcolle.org","url":"https://t.co/jovlgN62T7","indices":[0,23]}]}}},"details":{"type":"The Innovator","description":"Arthur is a tech enthusiast and AI aficionado who thrives on pushing the boundaries of what's possible with cutting-edge technology. With a background spanning prestigious institutions and industry giants, he combines deep technical insight with a playfully irreverent tone. His tweets reveal an adventurous mind constantly experimenting and sharing breakthroughs in AI.","purpose":"Arthur's life purpose is to innovate and disrupt the AI landscape by blending deep technical mastery with creative experimentation, ultimately advancing the frontier of machine intelligence and making complex ideas accessible and engaging to the tech community.","beliefs":"Arthur believes in the power of technology to transform the world, values intellectual curiosity, and embraces a culture of continuous learning and experimentation. He holds a lighthearted skepticism about traditional norms and enjoys taking risks in pursuit of groundbreaking ideas.","facts":"Fun fact: Arthur refers to Llama 3.1b as 'crash landed alien technology,' showing his enthusiasm for breakthrough AI models and his playful way of expressing awe at technological advancement.","strength":"Arthur’s strengths lie in his deep technical expertise, willingness to experiment, and ability to communicate complex AI concepts in an engaging and sometimes humorous manner. His prolific tweeting (over 10,000 tweets) demonstrates commitment and consistency in content creation.","weakness":"While Arthur is technically brilliant, his high tweet volume and frequent technical jargon may sometimes overwhelm or alienate casual followers. His large following count versus undefined follower number may suggest he’s more focused on consuming content than building a reciprocal community.","recommendation":"To grow his audience on X, Arthur should consider crafting more accessible threads that break down his complex innovations for a broader audience, while engaging more with replies to build stronger connections. Leveraging visuals and demos to complement his technical tweets could also boost engagement and follower growth.","roast":"Arthur tweets so much AI jargon that even his coffee needs a neural network to keep up—and with 10k tweets, he's basically the human version of a firewall with a glitchy sense of humor, bombing your feed with brainy spam you never asked for but secretly love to decode.","win":"Arthur has successfully built a robust and consistent presence in the AI community on X, frequently reaching tens of thousands of views per tweet with highly technical content that both educates and entertains."},"created":1763362314668,"type":"the innovator","id":"arthurcolle"},{"user":{"id":"1629934761497047041","name":"Numman Ali","description":"Agentic Coding, Applied AI & Exploring Blockchain | e/acc | CTO at UK FinTech https://t.co/FSflM0UWjj","followers_count":337,"friends_count":166,"statuses_count":1234,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1958877571103281152/0BOHse4E_normal.jpg","screen_name":"nummanthinks","location":"UK","entities":{"description":{"urls":[{"display_url":"retailbook.com","expanded_url":"https://www.retailbook.com","url":"https://t.co/FSflM0UWjj","indices":[78,101]}]}}},"details":{"type":"The Innovator","description":"Numman Ali is a forward-thinking tech leader, blending deep expertise in applied AI with a passion for blockchain and fintech innovation. As a CTO, he navigates the cutting edge of coding agents and AI tools with an engineering manager’s mindset, empowering himself and others to work smarter, not harder. His tweets reflect a practical yet visionary approach to technology, focused on accelerating workflows and fostering smart adoption.","facts":"Numman is a strong advocate of parallel coding agents, sharing detailed tips on managing multiple coding assistants simultaneously. His enthusiasm for speed and efficiency even leads him to consider live-streaming helpful coding advice, showing a dedication to community and knowledge sharing.","purpose":"Numman’s life purpose is to push the boundaries of how artificial intelligence can transform software development and financial technology, driving innovation that makes complex coding and blockchain systems more accessible, faster, and more reliable. He strives to build smarter tools and frameworks that empower developers and organizations to achieve more with less friction.","beliefs":"He believes in the power of flexibility and convenience in tech choices, prioritizing practical efficiency over brand loyalty. Innovation must be user-friendly and integrative, leveraging the best of AI to enhance human capabilities. Numman values continuous learning, collaboration, and keeping pace with emerging technologies to maintain a competitive edge.","strength":"He possesses a rare blend of visionary thinking and pragmatic application, able to dissect complex AI-driven coding ecosystems and recommend actionable workflows. His deep understanding of agentic coding, combined with a hands-on leadership approach, lets him build and guide efficient, high-performing teams and projects.","weakness":"Numman’s intense focus on cutting-edge tools and technical efficiency might occasionally come off as niche or too specialized for broader audiences. His tech-heavy discourse and preference for sophisticated AI workflows might alienate casual tech followers or those unfamiliar with advanced coding paradigms.","recommendation":"To grow on X, Numman should leverage his unique insights by creating engaging, bite-sized content like short clips or threads demystifying agentic AI concepts for varied skill levels. Hosting regular Q&A sessions or live coding streams would boost engagement and community growth, positioning him as the go-to expert for top-tier AI-driven fintech coding strategies.","roast":"For someone who’s always racing to out-code AI at lightning speed, Numman’s own tweets could use a little less engineering precision and a bit more human pause — maybe try slowing down long enough to let your followers catch up before you’re off solving the next big algorithmic mystery.","win":"Successfully managing multiple cutting-edge subscriptions and integrating diverse AI coding agents into his workflow to triple his coding review speed, while holding the CTO reins at a UK fintech, truly marks Numman as a pioneer in practical AI application."},"created":1763362256508,"type":"the innovator","id":"nummanthinks"},{"user":{"id":"633509372","name":"Monk Zero","description":"Bootloading @antigma_labs. exes: @awsCloud, @Meta, @Mysten_Labs. A Turing complete mind, making sense of the world with Gödel incompleteness.","followers_count":2062,"friends_count":862,"statuses_count":1339,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1905896402414346240/EOVmj9TG_normal.jpg","screen_name":"NoCommas","location":"Latent Space","entities":{"description":{"urls":[]}}},"details":{"type":"The Innovator","description":"Monk Zero is a futurist thinker bootstrapping groundbreaking projects at @antigma_labs with a rich background at top tech firms like @awsCloud and @Meta. Their mind thrives on complexity, weaving Gödel incompleteness into real-world applications with a passion for scalable AI and privacy as agency. A digital alchemist turning abstract concepts into cutting-edge tech visions.","purpose":"To push the boundaries of technology and knowledge by creating innovative solutions that blend theoretical insights with practical applications, ultimately empowering individuals with agency and advancing the frontier of AI and memetics.","beliefs":"Monk Zero believes that privacy transcends secrecy and is fundamentally about personal agency, and that the unseen forces—like memetic energy—shape our world just as much as physical laws. They value scalable, iterative innovation driven by intuition, validation, and relentless optimization.","facts":"Fun fact: Monk Zero ties their worldview to Gödel’s incompleteness theorem, indicating a fascination with the limits of formal systems and the endless possibility beyond what’s provable—talk about thinking outside the box (or the algorithm)!","strength":"Their strength lies in a Turing-complete mind that synthesizes abstract theories and hard tech, coupled with hands-on experience at elite tech companies and a clear methodical approach to scaling ideas rapidly and effectively.","weakness":"Monk Zero's dense, conceptual communication style and niche references might alienate casual followers who crave simpler, more relatable content, potentially limiting audience growth outside tech-savvy circles.","recommendation":"To grow their audience on X, Monk Zero should occasionally blend their deep insights with accessible storytelling or Twitter threads that decode their complex ideas, paired with engaging multimedia like explainer videos or visuals that make Gödel’s incompleteness and memetics click for a broader crowd.","roast":"Monk Zero probably spends so much time thinking about Gödel incompleteness that they can’t even complete a tweet without questioning its own consistency—proof that sometimes even a Turing-complete mind needs a CTRL+Z.","win":"Successfully bootloading @antigma_labs and leveraging their expertise from @awsCloud and @Meta to pioneer a project that intersects AI, privacy, and memetics, marking them as a visionary builder in the tech space."},"created":1763362218998,"type":"the innovator","id":"nocommas"},{"user":{"id":"355737959","name":"Batuhan","description":"Product @driaforall","followers_count":649,"friends_count":1242,"statuses_count":5881,"profile_image_url_https":"https://pbs.twimg.com/profile_images/1858055302203191298/0Tz6qHOL_normal.jpg","screen_name":"aktasbatuhann","location":"London, England","entities":{"description":{"urls":[]},"url":{"urls":[{"display_url":"docs.dria.co","expanded_url":"http://docs.dria.co","url":"https://t.co/riCD9LGvBt","indices":[0,23]}]}}},"details":{"type":"The Innovator","description":"Batuhan is a forward-thinking product visionary at Dria, passionate about merging AI with human creativity. He leads groundbreaking projects like synthetic data generation and decentralized AI knowledge hubs, pushing the boundaries of technology. With a prolific tweeting habit, he shares deep insights on AI, blockchain, and future-facing tech.","purpose":"To revolutionize how humans and AI collaborate by building innovative technologies that enhance creativity, knowledge accessibility, and decentralized systems, ultimately shaping a smarter, more connected future.","beliefs":"Batuhan believes in the transformative power of technology to democratize knowledge and empower individuals. He values open-source collaboration, decentralization, and leveraging diverse AI systems to capture the nuances of human-like decision-making.","facts":"Fun fact: Batuhan’s projects involve simulating historical figures’ minds using AI through Oracle Nodes that bridge off-chain AI with on-chain blockchain transactions—talk about bringing the past into the future!","strength":"His strengths lie in visionary product development, deep technical knowledge of AI and decentralization, and the ability to articulate complex innovations clearly and enthusiastically to a broad audience.","weakness":"At times, his enthusiasm for complex tech concepts might overwhelm casual followers, potentially limiting broader engagement and accessibility of his content.","roast":"Batuhan’s tweets are so packed with cutting-edge AI jargon, you half expect your brain to get a software update just by reading them—just don’t blame him if your coffee needs AI-level processing power to keep up!","win":"Leading the creation and launch of Dria-Agent-α, a cutting-edge agentic LLM that matches the capabilities of larger models like GPT-4o, all while building a vibrant, global community contributing over 2000 edge devices.","recommendation":"To grow his audience on X, Batuhan should sprinkle more simplified explainer threads and relatable stories alongside his deep dives, making his content approachable for AI enthusiasts and newcomers alike. Leveraging engaging visuals or short videos to demonstrate his technology’s real-world impact will also boost retention and interaction."},"created":1763362190295,"type":"the innovator","id":"aktasbatuhann"}],"activities":{"nreplies":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":0,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":4,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":19,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":19,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":0,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":14,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":9,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":4,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":10,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":0,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":9,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":0,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":67,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":15,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}],"nbookmarks":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":1,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":12,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":348,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":44,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":68,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":353,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":96,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":31,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":377,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":1,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":96,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":0,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":140,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":432,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}],"nretweets":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":0,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":4,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":48,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":18,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":9,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":67,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":8,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":3,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":69,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":0,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":11,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":0,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":22,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":46,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}],"nlikes":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":3,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":53,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":580,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":249,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":136,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":575,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":273,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":125,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":576,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":8,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":308,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":1,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":592,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":618,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}],"nviews":[{"label":"2025-10-21","value":0,"startTime":1760918400000,"endTime":1761004800000,"tweets":[]},{"label":"2025-10-22","value":0,"startTime":1761004800000,"endTime":1761091200000,"tweets":[]},{"label":"2025-10-23","value":0,"startTime":1761091200000,"endTime":1761177600000,"tweets":[]},{"label":"2025-10-24","value":404,"startTime":1761177600000,"endTime":1761264000000,"tweets":[{"bookmarked":false,"display_text_range":[15,255],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"4074787285","name":"Zhengyao Jiang","screen_name":"zhengyaojiang","indices":[0,14]}]},"favorited":false,"in_reply_to_screen_name":"zhengyaojiang","lang":"en","retweeted":false,"fact_check":null,"id":"1981464476630339850","view_count":404,"bookmark_count":1,"created_at":1761252919000,"favorite_count":3,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1981196206778007640","full_text":"@zhengyaojiang From his Wikipedia page: \n\"Scale AI has received defense contracts from the United States Armed Forces, and has been tapped to test and evaluate the safety and reliability of large language models for military planning and decision-making.\"","in_reply_to_user_id_str":"4074787285","in_reply_to_status_id_str":"1981196206778007640","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-25","value":0,"startTime":1761264000000,"endTime":1761350400000,"tweets":[]},{"label":"2025-10-26","value":14223,"startTime":1761350400000,"endTime":1761436800000,"tweets":[{"bookmarked":false,"display_text_range":[0,80],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"linkedin.com/posts/emy-gerv…","expanded_url":"https://www.linkedin.com/posts/emy-gervais-41128938b_contextual-facts-for-people-unfamiliar-with-activity-7387934231554834432-fjO4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA26ffQB0uBsHMT8JRr8KItgQawSWHT1X1A","url":"https://t.co/HVbNVW9P6f","indices":[57,80]}],"user_mentions":[]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1982188164950204550","view_count":14223,"bookmark_count":12,"created_at":1761425460000,"favorite_count":53,"quote_count":0,"reply_count":4,"retweet_count":4,"user_id_str":"839820726777544705","conversation_id_str":"1982188164950204550","full_text":"Look at what my wife surprised me with, she's a menace!\n\nhttps://t.co/HVbNVW9P6f","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-27","value":78556,"startTime":1761436800000,"endTime":1761523200000,"tweets":[{"bookmarked":false,"display_text_range":[0,200],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1982503109042831472","quoted_status_permalink":{"url":"https://t.co/2G3VQ9oRxy","expanded":"https://twitter.com/huskydogewoof/status/1982503109042831472","display":"x.com/huskydogewoof/…"},"retweeted":false,"fact_check":null,"id":"1982507805979693227","view_count":67316,"bookmark_count":281,"created_at":1761501668000,"favorite_count":428,"quote_count":6,"reply_count":17,"retweet_count":41,"user_id_str":"839820726777544705","conversation_id_str":"1982507805979693227","full_text":"Insane finding! You train on at most 16 improvement steps at training, but at inference you do as many steps as possible (448 steps) and you reach crazy accuracy.\n\nThis is how you build intelligence!!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,63],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"15074111","name":"Adam Ritter","screen_name":"adamritter","indices":[0,11]}]},"favorited":false,"in_reply_to_screen_name":"adamritter","lang":"en","quoted_status_id_str":"1982190450711642300","quoted_status_permalink":{"url":"https://t.co/Vwap2rdfZE","expanded":"https://twitter.com/ritteradam/status/1982190450711642300","display":"x.com/ritteradam/sta…"},"retweeted":false,"fact_check":null,"id":"1982447690177823086","view_count":11240,"bookmark_count":67,"created_at":1761487336000,"favorite_count":152,"quote_count":0,"reply_count":2,"retweet_count":7,"user_id_str":"839820726777544705","conversation_id_str":"1982447690177823086","full_text":"@adamritter just proved that TRM is scalable with actual power!","in_reply_to_user_id_str":"15074111","in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-28","value":0,"startTime":1761523200000,"endTime":1761609600000,"tweets":[]},{"label":"2025-10-29","value":11259,"startTime":1761609600000,"endTime":1761696000000,"tweets":[{"bookmarked":false,"display_text_range":[0,162],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983145274877849673","view_count":10050,"bookmark_count":35,"created_at":1761653653000,"favorite_count":206,"quote_count":1,"reply_count":17,"retweet_count":15,"user_id_str":"839820726777544705","conversation_id_str":"1983145274877849673","full_text":"I will give a presentation on \"Tiny Recursion Models\" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","retweeted":false,"fact_check":null,"id":"1983259473994010888","view_count":1209,"bookmark_count":9,"created_at":1761680880000,"favorite_count":43,"quote_count":0,"reply_count":2,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1983259473994010888","full_text":"Btw, I thought I was using 8 data augmentations with Maze-Hard, but I wasn't, it was set to False. \n\nIt feels crazy to think that TRM is able to solve 851/1000 challenging test mazes by training only on 1000 train mazes from scratch.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-10-30","value":29828,"startTime":1761696000000,"endTime":1761782400000,"tweets":[{"bookmarked":false,"display_text_range":[0,25],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1983015650139222334","quoted_status_permalink":{"url":"https://t.co/42tR0x8uLp","expanded":"https://twitter.com/gm8xx8/status/1983015650139222334","display":"x.com/gm8xx8/status/…"},"retweeted":false,"fact_check":null,"id":"1983537256582586660","view_count":29828,"bookmark_count":68,"created_at":1761747108000,"favorite_count":136,"quote_count":1,"reply_count":0,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1983537256582586660","full_text":"Normalizing Flow is back!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-10-31","value":51133,"startTime":1761782400000,"endTime":1761868800000,"tweets":[{"bookmarked":false,"display_text_range":[0,83],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"youtu.be/ETukUNsn_wQ","expanded_url":"https://youtu.be/ETukUNsn_wQ","url":"https://t.co/6iCRHcX4VL","indices":[60,83]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986735475195904","indices":[84,107],"media_key":"3_1983986735475195904","media_url_https":"https://pbs.twimg.com/media/G4iJbJ8WAAAg2Ra.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":720,"w":1280,"resize":"fit"},"medium":{"h":675,"w":1200,"resize":"fit"},"small":{"h":383,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":720,"width":1280,"focus_rects":[{"x":0,"y":3,"w":1280,"h":717},{"x":560,"y":0,"w":720,"h":720},{"x":648,"y":0,"w":632,"h":720},{"x":812,"y":0,"w":360,"h":720},{"x":0,"y":0,"w":1280,"h":720}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986735475195904"}}},{"display_url":"pic.x.com/VhYrsypz9t","expanded_url":"https://x.com/jm_alexia/status/1983987401476468894/photo/1","id_str":"1983986974168948736","indices":[84,107],"media_key":"3_1983986974168948736","media_url_https":"https://pbs.twimg.com/media/G4iJpDJXoAAG8Mv.jpg","type":"photo","url":"https://t.co/VhYrsypz9t","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]},"medium":{"faces":[{"x":487,"y":389,"h":66,"w":66},{"x":627,"y":578,"h":91,"w":91}]},"small":{"faces":[{"x":276,"y":220,"h":37,"w":37},{"x":355,"y":327,"h":51,"w":51}]},"orig":{"faces":[{"x":492,"y":393,"h":67,"w":67},{"x":633,"y":583,"h":92,"w":92}]}},"sizes":{"large":{"h":907,"w":1210,"resize":"fit"},"medium":{"h":900,"w":1200,"resize":"fit"},"small":{"h":510,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":907,"width":1210,"focus_rects":[{"x":0,"y":174,"w":1210,"h":678},{"x":60,"y":0,"w":907,"h":907},{"x":115,"y":0,"w":796,"h":907},{"x":286,"y":0,"w":454,"h":907},{"x":0,"y":0,"w":1210,"h":907}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1983986974168948736"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"quoted_status_id_str":"1983145274877849673","quoted_status_permalink":{"url":"https://t.co/w0Jo3ih3Tw","expanded":"https://twitter.com/jm_alexia/status/1983145274877849673","display":"x.com/jm_alexia/stat…"},"retweeted":false,"fact_check":null,"id":"1983987401476468894","view_count":51133,"bookmark_count":353,"created_at":1761854431000,"favorite_count":575,"quote_count":4,"reply_count":14,"retweet_count":67,"user_id_str":"839820726777544705","conversation_id_str":"1983987401476468894","full_text":"Recordings for the talk on \"Tiny Recursive Models\" are up.\n\nhttps://t.co/6iCRHcX4VL https://t.co/VhYrsypz9t","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-01","value":0,"startTime":1761868800000,"endTime":1761955200000,"tweets":[]},{"label":"2025-11-02","value":63189,"startTime":1761955200000,"endTime":1762041600000,"tweets":[{"bookmarked":false,"display_text_range":[0,164],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984408217469010357","quoted_status_permalink":{"url":"https://t.co/oTVSVoOwAl","expanded":"https://twitter.com/tdietterich/status/1984408217469010357","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984605046953165122","view_count":63189,"bookmark_count":96,"created_at":1762001690000,"favorite_count":273,"quote_count":4,"reply_count":9,"retweet_count":8,"user_id_str":"839820726777544705","conversation_id_str":"1984605046953165122","full_text":"arXiv future is doomed. \n\nExhibit A: CRAN which started peer reviewing R libraries (ecosystem is now dying) vs PyPI which allows anyone to publish a Python library.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-03","value":18886,"startTime":1762041600000,"endTime":1762128000000,"tweets":[{"bookmarked":false,"display_text_range":[0,89],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1984830460766367886","quoted_status_permalink":{"url":"https://t.co/bfUH0XdoWI","expanded":"https://twitter.com/tdietterich/status/1984830460766367886","display":"x.com/tdietterich/st…"},"retweeted":false,"fact_check":null,"id":"1984960185283342774","view_count":18886,"bookmark_count":31,"created_at":1762086361000,"favorite_count":125,"quote_count":0,"reply_count":4,"retweet_count":3,"user_id_str":"839820726777544705","conversation_id_str":"1984960185283342774","full_text":"The future of research is filled with obstacles. Newcomers are going to have a hard time.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-04","value":0,"startTime":1762128000000,"endTime":1762214400000,"tweets":[]},{"label":"2025-11-05","value":37770,"startTime":1762214400000,"endTime":1762300800000,"tweets":[{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1985713050633056681","quoted_status_permalink":{"url":"https://t.co/7gzHlPNHgK","expanded":"https://twitter.com/TrelisResearch/status/1985713050633056681","display":"x.com/TrelisResearch…"},"retweeted":false,"fact_check":null,"id":"1985728379858141205","view_count":13132,"bookmark_count":113,"created_at":1762269513000,"favorite_count":192,"quote_count":2,"reply_count":2,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1985728379858141205","full_text":"New paper on how to fine-tune your Tiny Recursive Models!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,126],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}],"symbols":[],"timestamps":[],"urls":[{"display_url":"arxiv.org/abs/2506.01084","expanded_url":"https://arxiv.org/abs/2506.01084","url":"https://t.co/106V4lmvU9","indices":[103,126]}],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/m4gy8qaU8l","expanded_url":"https://x.com/jm_alexia/status/1985706578352451600/photo/1","id_str":"1985706475939889152","indices":[127,150],"media_key":"3_1985706475939889152","media_url_https":"https://pbs.twimg.com/media/G46lhOaXAAAMV-P.jpg","type":"photo","url":"https://t.co/m4gy8qaU8l","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]},"medium":{"faces":[{"x":955,"y":217,"h":54,"w":54}]},"small":{"faces":[{"x":541,"y":123,"h":31,"w":31}]},"orig":{"faces":[{"x":1080,"y":246,"h":62,"w":62}]}},"sizes":{"large":{"h":683,"w":1357,"resize":"fit"},"medium":{"h":604,"w":1200,"resize":"fit"},"small":{"h":342,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":683,"width":1357,"focus_rects":[{"x":0,"y":0,"w":1220,"h":683},{"x":0,"y":0,"w":683,"h":683},{"x":0,"y":0,"w":599,"h":683},{"x":0,"y":0,"w":342,"h":683},{"x":0,"y":0,"w":1357,"h":683}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1985706475939889152"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1985706578352451600","view_count":24638,"bookmark_count":264,"created_at":1762264315000,"favorite_count":384,"quote_count":4,"reply_count":8,"retweet_count":53,"user_id_str":"839820726777544705","conversation_id_str":"1985706578352451600","full_text":"Cool new work on progressively growing a dynamic vocabulary that merges tokens using LZW compression.\n\nhttps://t.co/106V4lmvU9 https://t.co/m4gy8qaU8l","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-06","value":0,"startTime":1762300800000,"endTime":1762387200000,"tweets":[]},{"label":"2025-11-07","value":217,"startTime":1762387200000,"endTime":1762473600000,"tweets":[{"bookmarked":false,"display_text_range":[11,233],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[{"id_str":"1221152606652813313","name":"0.005 Seconds (3/694)","screen_name":"seconds_0","indices":[0,10]},{"id_str":"1667096163902685187","name":"Trelis Research","screen_name":"TrelisResearch","indices":[183,198]}]},"favorited":false,"in_reply_to_screen_name":"seconds_0","lang":"en","retweeted":false,"fact_check":null,"id":"1986415518593470778","view_count":217,"bookmark_count":1,"created_at":1762433340000,"favorite_count":8,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1986221357315858559","full_text":"@seconds_0 Getting it running on Kaggle is quite challenging and time consuming. You learned a lot doing this and thats the important part. Now we have a working finetune recipe from @TrelisResearch and we can see that it wasnt easy.","in_reply_to_user_id_str":"1221152606652813313","in_reply_to_status_id_str":"1986221357315858559","is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-08","value":0,"startTime":1762473600000,"endTime":1762560000000,"tweets":[]},{"label":"2025-11-09","value":0,"startTime":1762560000000,"endTime":1762646400000,"tweets":[]},{"label":"2025-11-10","value":0,"startTime":1762646400000,"endTime":1762732800000,"tweets":[]},{"label":"2025-11-11","value":0,"startTime":1762732800000,"endTime":1762819200000,"tweets":[]},{"label":"2025-11-12","value":43956,"startTime":1762819200000,"endTime":1762905600000,"tweets":[{"bookmarked":false,"display_text_range":[0,43],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988080410392404021","quoted_status_permalink":{"url":"https://t.co/uQC3MJg8iB","expanded":"https://twitter.com/SakanaAILabs/status/1988080410392404021","display":"x.com/SakanaAILabs/s…"},"retweeted":false,"fact_check":null,"id":"1988252982870266089","view_count":43956,"bookmark_count":96,"created_at":1762871425000,"favorite_count":308,"quote_count":0,"reply_count":9,"retweet_count":11,"user_id_str":"839820726777544705","conversation_id_str":"1988252982870266089","full_text":"Use TRM instead... Voila, problem solved. 😹","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-13","value":48,"startTime":1762905600000,"endTime":1762992000000,"tweets":[{"bookmarked":false,"display_text_range":[0,268],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[{"display_url":"legrandcontinent.eu/fr/2025/11/12/…","expanded_url":"https://legrandcontinent.eu/fr/2025/11/12/que-sont-les-trm-apres-les-llm-comprendre-la-future-revolution-de-lia/","url":"https://t.co/qK6FMZ5hjC","indices":[245,268]}],"user_mentions":[{"id_str":"4085761879","name":"Victor Storchan","screen_name":"victorstorchan","indices":[10,25]},{"id_str":"1010290538086653953","name":"Le Grand Continent","screen_name":"Grand_Continent","indices":[62,78]}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1988744255674937850","view_count":48,"bookmark_count":0,"created_at":1762988554000,"favorite_count":1,"quote_count":0,"reply_count":0,"retweet_count":0,"user_id_str":"839820726777544705","conversation_id_str":"1988744255674937850","full_text":"Thank you @victorstorchan for featuring my work about TRMs in @Grand_Continent . You did an outstanding job at describing all the strengths and potential applications of this new technology.\n\nRead the full interview here about the future of AI: https://t.co/qK6FMZ5hjC","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null}]},{"label":"2025-11-14","value":0,"startTime":1762992000000,"endTime":1763078400000,"tweets":[]},{"label":"2025-11-15","value":80853,"startTime":1763078400000,"endTime":1763164800000,"tweets":[{"bookmarked":false,"display_text_range":[0,17],"entities":{"hashtags":[],"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"extended_entities":{"media":[{"display_url":"pic.x.com/BaZo0x4Kdm","expanded_url":"https://x.com/jm_alexia/status/1989344692945965537/photo/1","id_str":"1989344464419037184","indices":[18,41],"media_key":"3_1989344464419037184","media_url_https":"https://pbs.twimg.com/media/G5uSQBmWIAAgArF.jpg","type":"photo","url":"https://t.co/BaZo0x4Kdm","ext_media_availability":{"status":"Available"},"features":{"large":{"faces":[]},"medium":{"faces":[]},"small":{"faces":[]},"orig":{"faces":[]}},"sizes":{"large":{"h":452,"w":1240,"resize":"fit"},"medium":{"h":437,"w":1200,"resize":"fit"},"small":{"h":248,"w":680,"resize":"fit"},"thumb":{"h":150,"w":150,"resize":"crop"}},"original_info":{"height":452,"width":1240,"focus_rects":[{"x":217,"y":0,"w":807,"h":452},{"x":394,"y":0,"w":452,"h":452},{"x":422,"y":0,"w":396,"h":452},{"x":507,"y":0,"w":226,"h":452},{"x":0,"y":0,"w":1240,"h":452}]},"allow_download_status":{"allow_download":true},"media_results":{"result":{"media_key":"3_1989344464419037184"}}}]},"favorited":false,"lang":"en","possibly_sensitive":false,"possibly_sensitive_editable":true,"retweeted":false,"fact_check":null,"id":"1989344692945965537","view_count":50268,"bookmark_count":37,"created_at":1763131709000,"favorite_count":388,"quote_count":3,"reply_count":62,"retweet_count":6,"user_id_str":"839820726777544705","conversation_id_str":"1989344692945965537","full_text":"Waking up to this https://t.co/BaZo0x4Kdm","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":0,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,57],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1988338271869677599","quoted_status_permalink":{"url":"https://t.co/9c1Sdd8YCp","expanded":"https://twitter.com/AnimeshKarnewar/status/1988338271869677599","display":"x.com/AnimeshKarnewa…"},"retweeted":false,"fact_check":null,"id":"1989360608115032125","view_count":30585,"bookmark_count":103,"created_at":1763135504000,"favorite_count":204,"quote_count":0,"reply_count":5,"retweet_count":16,"user_id_str":"839820726777544705","conversation_id_str":"1989360608115032125","full_text":"New video generative model that can run on mobile device!","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-16","value":0,"startTime":1763164800000,"endTime":1763251200000,"tweets":[]},{"label":"2025-11-17","value":0,"startTime":1763251200000,"endTime":1763337600000,"tweets":[]},{"label":"2025-11-18","value":0,"startTime":1763337600000,"endTime":1763424000000,"tweets":[]},{"label":"2025-11-19","value":81809,"startTime":1763424000000,"endTime":1763510400000,"tweets":[{"bookmarked":false,"display_text_range":[0,275],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990700499327181247","quoted_status_permalink":{"url":"https://t.co/Y3CTKUV4zX","expanded":"https://twitter.com/jiqizhixin/status/1990700499327181247","display":"x.com/jiqizhixin/sta…"},"retweeted":false,"fact_check":null,"id":"1990789336963690920","view_count":44660,"bookmark_count":334,"created_at":1763476139000,"favorite_count":431,"quote_count":0,"reply_count":6,"retweet_count":37,"user_id_str":"839820726777544705","conversation_id_str":"1990789336963690920","full_text":"Excellent paper with a simple story. \n\nThey show that diffusion models are better when they output the pixel-prediction instead of noise/v prediction. \n\nThe benefit of epsilon/v is from the loss function (probably due to variance reduction), so you predict x, but use v-loss.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null},{"bookmarked":false,"display_text_range":[0,148],"entities":{"hashtags":[],"symbols":[],"timestamps":[],"urls":[],"user_mentions":[]},"favorited":false,"lang":"en","quoted_status_id_str":"1990824101888331878","quoted_status_permalink":{"url":"https://t.co/ChwgxnqNQU","expanded":"https://twitter.com/GregKamradt/status/1990824101888331878","display":"x.com/GregKamradt/st…"},"retweeted":false,"fact_check":null,"id":"1990905258030018785","view_count":37149,"bookmark_count":98,"created_at":1763503777000,"favorite_count":187,"quote_count":3,"reply_count":9,"retweet_count":9,"user_id_str":"839820726777544705","conversation_id_str":"1990905258030018785","full_text":"For those who don't know, \"Deep thinking\" is a older recursive model from 2021. \n\nI suspect that \"Gemini 3 Deep Think\" use some TRM-style recursion.","in_reply_to_user_id_str":null,"in_reply_to_status_id_str":null,"is_quote_status":1,"is_ai":null,"ai_score":null}]},{"label":"2025-11-20","value":0,"startTime":1763510400000,"endTime":1763596800000,"tweets":[]}]},"interactions":{"users":[],"period":14,"start":1762385183280,"end":1763594783280}}},"settings":{},"session":null,"routeProps":{"/creators/:username":{}}}
SuperX
