Caleb
3,881 posts
Caleb
@calebfahlgren
- WOW! Facebook drops a dataset of 1M+ reasoning traces 🤯 It consists of high-quality challenging reasoning questions backtranslated from pretraining corpora DCLM and FineMath. The dataset includes the extracted reference answer from the document! Amazing @facebook 👏
- qwen-2.5 coder is absolutely cracked when you give it a code interpreter 😈 and it's only a tiny 1.5B model running in the browser!
00:00 - Deepseek Artifacts 🖼️ Let the most powerful open model in the world create landing pages for you 👀 • Powered by Deepseek V3 🐋 • Very strong React / Tailwind ability 💪 Come play with it and help build the largest, high quality frontend code dataset on @huggingface
00:00 - Qwen-2.5 on WebGPU 🏎️ • 42 tok/sec for Qwen2.5-Coder-1.5B on Mac ⚡ • Powered by MLC WebLLM and WebGPU 🔥 Watch Qwen2.5-Coder-1.5B build a website entirely in the browser!
00:00 - Released a free tool on ChatDB: Parquet AI Query parquet files with natural language in the browser. ◆ Powered by @duckdb in the browser ◆ LLM from @GroqInc and Llama-3-70B Here's me querying the capybara-dpo dataset from @huggingface
00:00 - My favorite thing about working at the @huggingface Paris office: You can ship a feature into prod and then go hit deadlifts one floor below 💪
- Well this did well 👀 Dataset is now at 35k unique generations of different react apps 🔥 • 37.3K - Deepseek V3 🐋 • 5.2k - Qwen2.5-Coder 32B and growing 🦦 • 41 - Llama 3.1 405B 🦙 Switched out Deepseek V3 with Qwen2.5-Coder-32B on @huggingface Inference API to save on myDeepseek Artifacts 🖼️ Let the most powerful open model in the world create landing pages for you 👀 • Powered by Deepseek V3 🐋 • Very strong React / Tailwind ability 💪 Come play with it and help build the largest, high quality frontend code dataset on @huggingface00:00
00:00- Reasoning datasets are flooding the top 30 trending datasets on Hugging Face. I created a collection of them here huggingface.co/collections/cf…
- Excited to release Natural-SQL-7B! A new, very strong Text to SQL model that is my best fine tune yet. You can see how it does on the SQL-Eval benchmark by @defogdata huggingface.co/chatdb/natural…
