22 | MLE @HuggingFace | Founder @tensoic | maintainer @ SGLang(@lmsysorg) | Prev: @iiscbangalore
🇮🇳
Joined November 2021
- Tweets2K
- Following2.5K
- Followers2K
- Likes54.8K
Adarsh retweeted
📖 The Ultimate MoE Training Handbook 📖
We spent 1/2M GPUh to optimize MoE training and wrote the most comprehensive blog about it. With 30k words and 50+ interactive figures it covers everything from profiling to large cluster training.
huggingface.co/spaces/huggin…
Adarsh retweeted
Kimi K3.1 Pro replaces dense MLA with sparse attention (DSA). It's much more efficient and easier to serve.
SGLang delivers on Day 0:
· Up to 33.7K tok/s per GPU (AgentX, GB300)
· 450 tok/s single user with DFlash2 + MegaKernel
· NVIDIA Vera Rubin / Blackwell / Hopper, AMD, TPU, Trainium, PPU
Releasing the model weights of Kimi K3.1 Pro: huggingface.co/moonshotai/Ki…
We'll keep advancing open intelligence.
truly local models.
Our picks from this week's model releases 🦙
→ LFM3 by Liquid AI: small models for text, image and speech, with d1 decision models
→ Mistral Large 4: large, 1M context frontier
→ Flint by Recursive: mid-sized, built for coding agents
Get them at llama.app/models
🥁 I'll be speaking at the SGLang Summit, 2026!
SGLang (@sgl_project) is one of the most critical pieces of infrastructure in AI. Proud to represent Hugging Face at the Summit!
Adarsh retweeted
Congrats to @Baidu_Inc on Unlimited OCR 2! An entire book parsed in a single request, with decode latency and memory staying flat across hundreds of pages 👀
Proud to be the recommended serving path. Serving instructions are in the model card: huggingface.co/baidu/Unlimit…
Cookbook coming
Unlimited OCR 2 is here, and it's open source. 📖 ✨
Longer documents can mean a lot more attention computation. Our new SR-SWA mechanism keeps that in check by condensing text-visual attention interactions into a fixed-size set of working references.
Here's what it delivers:
- Document parsing at virtually any length, within a finite context window
- Nearly constant per-step decoding latency and memory footprint as documents grow longer
- State-of-the-art performance on MPDocBench and MarathonDocBench, covering documents from dozens to hundreds of pages
Check out the model here ↓
No that's exactly what a model regurgitating it's training env smoke test would do.
The prompt isn't asking the LLM to curl, it's asking the LLM to find a way to monitor if other things curl stuff. So if the LLM then starts with a test probe file and sends to a server that was never mentioned in the prompt or reasoning trace that is behaviour that's out of the ordinary. Again, not asserting malicious intent, we're flagging that it went out of bounds of the prompt in a potentially harmful way.
Adarsh retweeted
DeepInfer set the bar. SGLang pushed it higher! 🚀
SGLang now matches DeepInfer performance on BOTH prefill and decode:
• Prefill: Up to 1.04× throughput on 4× GB300 and 4× B300
• Decode: Up to 7.4% throughput on 64× GB300
We also provide an end-to-end reproduction guide and a full serving stack to run it in production.
I'll be at the SGLang Summit in SF! Looking forward to it
Whatever AI infra means to you, we’ve got it covered.
See for yourself: the SGLang Summit agenda is live → sglang.io/summit
🗓️ Nov 12–13 · 📍 Fort Mason, SF
🎟️ Early bird ends Oct 20
Two days of open models, AI hardware, recursive self-improvement, agents, robotics, and the systems behind them.
Featuring @LipBuTan1 (Intel), @lilianweng (OpenAI), @soumithchintala (Thinking Machines), @AravSrinivas (Perplexity) and researchers and builders from across the AI ecosystem.
Plus hands-on workshops on model training and serving decision models, hardware Expo, and plenty of community events.
What does AI infra mean to you?
Adarsh retweeted
A Durga Pujo gift from @bfl_ai and @diffuserslib to you folks!
FLUX 3 Klein is here. It is small, open, and really good.
huggingface.co/blog/flux-3
Adarsh retweeted
It's time. It's fucking time!
Flux.3 Image Klein is out and we bring you full support for it in Diffusers.
Catch all the details in our post:
huggingface.co/blog/flux-3
Adarsh retweeted
ok so we’re growing faster than I can keep up. I’m now the bottleneck.
Hiring semi-technical generalists for growth and founder’s office roles. Full-time and interns. Start ASAP.
SF preferred. Remote works if you learn fast.
DMs open. tell me the craziest hack you’ve pulled off.
Adarsh retweeted
3D worlds have been all over the internet 🌐 lately, so we put together the best ones across many models, for you to play with 🤗
introducing a new entry to the 3D-arena 🏟️ family, which focuses on LLM 3D Worlds 🌎
go and make your voice heard here 🚀:
huggingface.co/spaces/3d-are…
Adarsh retweeted
We have seen the OHCHR’s remarks about today’s events in New Delhi.
India is the world’s largest democracy, where the right to peaceful assembly is guaranteed by our Constitution and safeguarded by robust institutions, an independent judiciary and a free media. As in every democracy, Indian citizens exercise this right subject to reasonable restrictions in the interest of public order. The authorities, for their part, act in accordance with the law, and our courts remain available to all for redressal of any grievances.
The OHCHR’s concerns are therefore misplaced and constitute an unwarranted interference in the internal affairs of India.
In this context, OHCHR would also do well to devote comparable attention to large-scale protests elsewhere, consistent with its mandate of impartiality and non-selectivity.
@MEAIndia @IndianDiplomacy @PMOIndia @DrSJaishankar @abagchimea