Asynchronous GRPO with LoRA on Hugging Face Jobs: one bucket, one proxy, no NCCL
Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL
What happened
TRL v1.14's AsyncGRPOTrainer can now train and sync only LoRA adapters to vLLM. Rank-1 adapters are just a few megabytes and can be transferred through a Storage Bucket mounted across multiple Hugging Face Jobs, without NCCL.
Hugging FaceOpen sourceDeployment engineeringData and trainingTutorials#Tutorials and practice#Data and training#Deployment and engineering#Open-source ecosystem#Hugging Face
Keep reading
Latest daily · Issue 24NVIDIA and AI4Bharat Introduce Nemotron-pretraining-Indic-v1 and an Open Data Generation PipelineTrending now
- 1Meta Releases Muse Spark Open With Mac and Cloud Support, Topping US Open-Weight Index
- 2Google Launches Gemini 4 Argon Globally, Expands Paid Access as Benchmark Results Vary
- 3OpenAI Updates GPT-6 Astra Pro in ChatGPT With Parallel Agents and Faster Responses
- 4Meta’s New Business Terms Raise Dispute Over Website Crawling and AI Training Rights
Tomorrow at 08:00, a briefing like this arrives in your inbox.
The same kind of briefing as the one above: a few stories worth reading. Free. Mail starts after you confirm, and every letter has a one-click unsubscribe. Read a past issue · 2026-10-17