Hugging Face's logo
Activity Feed

AI & ML interests

None defined yet.

Recent Activity
ymoslem  authored a paper 9 days ago
Cluster, Route, Escalate: Cascaded Framework for Cost-Aware LLM Serving
ymoslem  submitted a paper 9 days ago
Cluster, Route, Escalate: Cascaded Framework for Cost-Aware LLM Serving
szymanowiczs  authored a paper 11 days ago
Splatter Image: Ultra-Fast Single-View 3D Reconstruction
View all activity
sergiopaniego 
posted an update about 5 hours ago
view post
Post
265
Frontier models use distillation as a step of their post-training pipelines.

In 2026 it has three jobs: compress a big model into a small one, merge RL experts into a single model, and let a model teach itself.

I wrote up which frontier models use each one and how: https://huggingface.co/blog/sergiopaniego/distillation-2026

It pairs with Class 2 of the Training an Agent series Ben and I are doing, where we teach these techniques hands-on with TRL!
yuntian-deng 
authored a paper 3 days ago
yuntian-deng 
submitted a paper to Daily Papers 6 days ago
IliaLarchenko 
submitted a paper to Daily Papers 10 days ago
IliaLarchenko 
authored a paper 11 days ago
IliaLarchenko 
posted an update 12 days ago
view post
Post
155
I placed 🥈 2nd in the LeHome Challenge (ICRA 2026), and 🥇 1st of 62 teams in the first simulation round. Now I'm open-sourcing the full solution — code, tech report, and final weights.

The task: teach a cheap two-armed robot (SO-ARM101) to fold 4 garment types — long/short tops and pants. Garment category is hidden at eval. Round 1 in sim (auto-scored), round 2 on a real robot (jury-scored).

I trained a VLA policy with an RL loop on top. The key ideas:

🧠 The policy is its own value function. From the same forward pass that picks the next action chunk, cheap heads predict success probability, task completion %, garment type, and future keypoint distances + a Q-residual. Those become the advantage signal for RL — no separate critic.

🔁 A fully asynchronous RL loop coordinated only through the HF Hub: 1 trainer (H200) ships a fresh checkpoint ~every 40 min while N rollout workers (and a human doing teleop / DAgger corrections) collect data in parallel. Nobody waits — it uses the off-policy nature of the loop to the fullest.

📈 Binary success is too sparse, so I densify it into per-frame advantage via GAE — from objective keypoint checkpoints, the success-probability value baseline, and completion %.

🎛️ The RL combines AWR + RECAP. I also tune the inference knobs — execution length, playback speed, inpainting overlap, CFG scale, best-of-N — with a per-parameter Thompson-sampling bandit folded into rollout collection.

🔧 Round 2: with only ~1 week and no access to the eval robot — so the pipeline was sim → my robot → their robot, leaning on heavy augmentation to make the policy more robust.

📝 Blog: https://ilialarchenko.com/projects/lehome2026
📄 Tech report: Learning to Fold: prizewinning solution at LeHome Challenge 2026 (1st place online, 2nd offline) (2606.27163)
🔧 Code: https://github.com/IliaLarchenko/lehome_solution
🤗 Sim policy: IliaLarchenko/lehome_sim
🤗 Real policy: IliaLarchenko/lehome_real
🌐 Challenge: https://lehome-challenge.com/
sergiopaniego 
posted an update 12 days ago
eienmojiki 
posted an update 13 days ago
tianweiy 
authored a paper 13 days ago
ajibawa-2023 
posted an update 18 days ago
view post
Post
6856
Shell-Code-Large
Dataset: ajibawa-2023/Shell-Code-Large

Shell-Code-Large is a large-scale corpus of Shell scripting source code comprising approximately 640,000 code samples stored in JSON Lines (.jsonl) format. The dataset is designed to support research in large language model (LLM) pretraining, code intelligence, DevOps automation, cloud infrastructure engineering, system administration, and software engineering automation.

By providing a high-volume, language-specific corpus focused exclusively on Shell scripting, Shell-Code-Large enables systematic experimentation in automation workflows, deployment pipelines, infrastructure management, and command-line tooling. These domains remain foundational to Linux systems, cloud-native platforms, CI/CD environments, and modern DevOps practices.

Shell-Code-Large addresses the need for a dedicated Shell-focused dataset at substantial scale, enabling targeted research into scripting patterns, command composition, workflow orchestration, infrastructure automation, and operational engineering practices
sergiopaniego 
posted an update 19 days ago
sergiopaniego 
posted an update 21 days ago
view post
Post
280
GLM-5.2 is open and comes with competitive performance against opus 4.8

day-0 in transformers + vllm + sglang, mit license 🤗

on the post-training side: critic-based ppo for variable-length agentic rollouts (ppo is back!) + an online anti-reward-hacking module that feeds the agent dummy info when it tries to cheat
mapooon 
authored a paper 23 days ago
zhiqiulin 
posted an update 29 days ago
view post
Post
121
🚀 VQAScore now supports text-to-video evaluation!

VQAScore scores how well a generated image or video matches a prompt by asking a VLM "does this show {prompt}?" and using P(Yes). It became a go-to evaluation metric and reward model for image generation (2M+ downloads), and we just added text-to-video support across 20+ VLMs (GPT, Gemini, Qwen). Free and open-source, and it keeps improving as VLMs improve.

💻 Code: https://github.com/linzhiqiu/t2v_metrics
📄 Paper: https://arxiv.org/abs/2404.01291
🧵 Launch thread + demo video: https://x.com/ZhiqiuLin/status/2064316582461841499
  • 1 reply
·
sergiopaniego 
posted an update about 1 month ago
view post
Post
3921
OpenEnv has a new home: github.com/huggingface/OpenEnv

Starting today, it's coordinated by a committee that includes Meta-PyTorch, Reflection, Unsloth, Modal, Prime Intellect, Nvidia, Mercor, Fleet AI, and Hugging Face

frontier labs train their models and their harnesses together. Claude knows Claude Code. GPT-5.5 knows Codex. that's not an accident, it's training. open-source models deserve the same magic, but pulling that off requires infrastructure that belongs to everyone, not one lab

OpenEnv is that layer. one api, any harness, any trainer, any environment

Rewards and training loops stay in TRL, Unsloth, wherever you already work. OpenEnv is the socket they all plug into

Get involved!

Full announcement: https://huggingface.co/blog/openenv-agentic-rl
yuntian-deng 
authored 3 papers about 1 month ago
sergiopaniego 
posted an update about 1 month ago
view post
Post
324
Frontier agents are this good partly because the model was trained inside the very harness it ships with.

NVIDIA's new paper "Polar: Agentic RL on Any Harness at Scale" brings that recipe to the open: it turns coding harnesses like Codex, Claude Code, Qwen Code or Pi into RL training environments without touching their internals.

The core idea: every agent, however complex or closed, talks to a model through an API, so they put a proxy there. The harness runs exactly like in production while the proxy records prompts, sampled token ids and logprobs. Trajectories get rebuilt outside, token faithful, so gradients hit the exact tokens the policy sampled.

The gains are consistent across all four harnesses. Same Qwen3.5-4B, plain GRPO, evaluated on SWE-Bench Verified:

Codex 3.8 → 26.4 (+22.6)
Claude Code 29.8 → 34.6 (+4.8)
Qwen Code 34.6 → 35.2 (+0.6)
Pi 34.2 → 40.4 (+6.2)

The biggest gains appear on unfamiliar execution paths, Codex being the clearest case. The takeaway: you are not just training a model, you are training the model + harness system.

Two engineering pieces make it work at scale. Async worker pools isolate container boots (CPU), agent execution (GPU) and long tail test runs, so slow runtimes never block the GPUs. And prefix merging stitches hundreds of captured API calls back into contiguous traces: 5.4x faster trainer updates and rollout GPUs at 88% utilization.

It also doubles as an SFT data factory: 504 test verified agent traces from a 122B teacher, multi-turn conversations averaging 104 messages each, coming to the Hub under Apache 2.0 (release pending review).

Paper authors: Binfeng Xu, Hao Zhang, Shaokun Zhang, Songyang Han, Mingjie Liu, Jian Hu, Shizhe Diao, Zhenghui Jin, Yunheng Zou, Michael Demoret, Jan Kautz and Yi Dong.

> Paper: Polar: Agentic RL on Any Harness at Scale (2605.24220)
> Code: https://github.com/NVIDIA-NeMo/ProRL-Agent-Server
> Training data: NovaSky-AI/SkyRL-v0-293-data
wangfuyun 
authored a paper about 1 month ago