Shanshan Wang

6 months ago we were at like 1.5k stars and pretty much the only ones doing this. now there’s a lot more players and the space is getting crowded, which honestly is a good sign. bring it on👻

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 ↓

We finished the Training Agents series. Six live sessions over six months, from evaluating agents to training them inside real environments. All of it is on the Hugging Face YouTube channel and all of the code is open. Here's what we did and who made it happen: 1. Agentic Evaluations WorkshopWhere agent evals actually stand, and why benchmark scores don't match what people see in use. With Avijit Ghosh and Nathan Habib (Hugging Face), Arvind Narayanan (Princeton), Pierre Andrews (Meta), J.J. Allaire (UK AI Security Institute) and Mahesh Sathiamoorthy (Bespoke Labs). 2. RL for Agents Workshop Environments, rollouts, reward design and the inference bottlenecks that appear when you move from RL for LLMs to RL for agents. With Lewis Tunstall (Hugging Face), Will Brown (Prime Intellect), Ofir Press (Princeton) and Alex Zhang (MIT CSAIL). 3. Training Agents 1: SFT on agent traces Public coding-agent traces turned into prompt/completion data, a TRL + LoRA fine-tune on Hugging Face Jobs, metrics in Trackio, and an honest look at what the first eval numbers can and cannot tell you. Joined by Sergio Paniego and Quentin Gallouédec. 4. Training Agents 2: Distillation Off-policy, on-policy and self-distillation for moving capability from a teacher into a smaller coding agent. 5. Training Agents 3: Reinforcement learning GRPO after SFT: group sampling, verifiable reward functions, reading the reward/KL/length curves, and three experiments, one of them with a deliberately gameable reward so we could watch the hacking happen. 6. Training Agents 4: From reward functions to environments The reward stops being a function and becomes a place the agent acts in. We walked the reset()/step() contract from Gym to LLM agents, built an OpenEnv environment and pushed it to the Hub, plugged it into TRL's GRPOTrainer, then trained a real coding agent (OpenCode) through Harbor with AsyncGRPOTrainer on Hugging Face sandboxes. The series has passed 300k views. Thank you to every speaker, to the TRL team, and to everyone who showed up live with questions. Playlist: youtube.com/playlist?list=PL…

Ran the best frontier models on industrial inspection notes and pick labels. Blindly but with full labels list. It’s already like an open book test but they still missed a lot. Reading and understanding the input text is not a problem, but the right label depends on conventions only the domain experts know. And even worse, the conventions aren’t always clear or consistently documented. So It’s always how fast you can consolidate and inject this knowledge, still a data problem at the end.

One of the first features we built into Terminal-Bench (long before Harbor existed) was --livestream to watch the tmux session. Glad to see @kobe0938 cooking up the better, more generalized version for Harbor :)