Keep on to blur preview images; turn off to show them clearly
微博把罗伯特开源了？ 微博AI刚刚发布了首个大模型！ VibeThinker-1.5B！ 最大的亮点是，这个模型AIME25数学能力测试分数比DeepSeek-R1(1月份版本)分数还高！ 以及值得一提的是，训练(微调)仅花费了7800美元。另外，这个模型是基于Qwen2.5-Math-1.5B 微调的。 目前官方没有部署space，测试需要把模型下载到本地运行，我正在弄一个 huggingfaceSpace 然后稍后给大家测一下这个模型的数学能力真的有这么强吗？ 一般要是模型自己说自己有多猛有多猛那肯定是要让我看♂看♂的。稍后放出测试结果。 模型地址：
基础数据
RT @samhogan: Trust me when I say Egocentric is a fantastic choice of name for an open source dataset
Co-founder & CEO @HuggingFace 🤗, the open and collaborative platform for AI builders
RT @feltanimalworld: 当年明月悟到几个高认知人士在20-40岁必经的阶段： 1）意识到自己的大脑是宇宙频率的容器。不是用来“计算”，而是用来“接收”。 2）看透孤独只是表像，通过阅读和反思带来的能量增长，远超低效社交的无数倍。…
Solo Developer ｜ Growth Coach｜Helping creators build their personal brand on X 公众号：PandaTalk8
RT @randall_balestr: LeJEPA: a novel pretraining paradigm free of the (many) heuristics we relied on (stop-grad, teacher, ...) - 60+ arch.,…
Professor at NYU. Chief AI Scientist at Meta. Researcher in AI, Machine Learning, Robotics, etc. ACM Turing Award Laureate.
RT @11AMdotclub: TODAY. LIQUIDITY DISTRIBUTION PROTOCOL. SOLANA GRAND PRIX. BOND WITH CREATORS. 11AM ET w/ @SuperChad15 of @turtledotxyz…
CEO @dotfunlabs - crypto games that drive culture ⿻ Crew Chief @ Solana Grand Prix / @racer_dot_fun
CodeZero is our newest environment on RL Swarm. Your node runs as a Solver: training and evaluating coding agents as part of the swarm. Plug in and help the network learn.
COO @GensynAI / ex @a16z @Cravath