Mishig Davaadorj
@mishig25
artificial intelligence @huggingface e/acc 🇲🇳🇫🇷
Dropped the Virtual Cell Challenge Primer on HF. We are shipping transformers support for STATE (the SOTA model for predicting perturbation response) very soon!
"narrative violation"
models like Kimi, DeepSeek and Qwen will cost the closed AI labs BILLIONS of dollars. that's why nobody is talking about them. despite these LLMs absolutely crushing all of the benchmarks. Claude 4 Opus is literally *100x* more expensive than Kimi K2 yet both models have…
If you are in field "xyz", make sure you are the best person in AI for "xyz"
🚨 Alert: Qwen3-Coder is now available in DeepSite. Probably one of the best way to try it with zero friction 🔥
>>> Qwen3-Coder is here! ✅ We’re releasing Qwen3-Coder-480B-A35B-Instruct, our most powerful open agentic code model to date. This 480B-parameter Mixture-of-Experts model (35B active) natively supports 256K context and scales to 1M context with extrapolation. It achieves…
OmniSVG demo & weights finally dropped on Hugging Face🔥🔥 ✨ end-to-end multimodal SVG generator ✨ leverages pre-trained VLMs ✨ works all the way from simple icons to intricate anime characters
🚨 Olympiad math + AI: We ran Google’s Gemini 2.5 Pro on the fresh IMO 2025 problems. With careful prompting and pipeline design, it solved 5 out of 6 — remarkable for tasks demanding deep insight and creativity. The model could win gold! 🥇 #AI #Math #LLMs #IMO2025
Inference compute vastly outweighs training compute in the long run. So you want inference to be cheap: * train longer to shrink the final model * use less inference-time scaling (i.e. no "thinking" as shown in new qwen)
Any way I can try new Qwen with Claude Code?
Qwen COOKED - beats Kimi K2 and competitive to Claude Opus 4 at 25% total parameters 🤯
Pretty crazy that this can be done: image+text prompt -> RGBD videos with predicted actions. Great chat with Haoyu!
Ep#21 with Haoyu Zhen on TesserAct: Learning 4D Embodied World Models tesseractworld.github.io Co-hosted by @micoolcho @chris_j_paxton
The interior design of this bar in New Jersey is straight up retro heaven
LTX Video 0.9.8 is here 🔥🔥 ✨ 2 new distilled checkpoints - 2B & 13B + IC LoRA for detail enhancement ✨ improved prompt adherence & detail generation ✨ blazing fast 💨 Try the new 0.9.8 13B 🤖👇