# Hugging Face — Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL

- Company: Hugging Face (huggingface.co)
- Announced: 2026-09-10
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/asyncgrpo-lora-hfjobs
- Record: https://forck.live/items/10541-async-grpo-with-lora-across-hf-jobs-a-bucket-a-proxy-and-no-nccl
- Subject: Platform

Hugging Face released LoRA support in TRL's AsyncGRPOTrainer (v1.14), enabling training of LoRA adapters that sync only the adapter weights to vLLM inference servers rather than full model weights. The implementation allows trainer and inference to run on separate Hugging Face Jobs using Storage Buckets for adapter synchronization and a proxy server for routing and broadcasting, reducing training time from 3 hours 27 minutes to 53 minutes for 500 steps in a real-world project.

## Evidence

Verbatim from https://huggingface.co/blog/asyncgrpo-lora-hfjobs:

> LoRA support recently landed in TRL's AsyncGRPOTrainer with PR #7017, and ships with TRL v1.14. The asynchronous trainer can now train an adapter instead of the full model, and it syncs only the LoRA adapter to vLLM.

---

Record: https://forck.live/items/10541-async-grpo-with-lora-across-hf-jobs-a-bucket-a-proxy-and-no-nccl
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md