Dirhousssi Amine(@DirhousssiAmine):We ran async GRPO across HF Jobs with the trainer and vLLM on separate machines, no NCCL, and no shared disk. The adapter is mounted at the same path in every Job. LoRA training has landed in TRL's AsyncGRPOTrainer 🎉🎉🎉 The trainer now syncs only the adapter to vLLM ~few MB for a 1.5B model. That changes what the setup can look like. We ran async GRPO across HF Jobs with the trainer and vLLM on separate machines, no NCCL, no shared disk. The adapter travels through a 🪣 mounted on the same path in every Job. Blogpost 🧵

Dirhousssi Amine

@DirhousssiAmine

🇲🇦 ML engineer - post training team @huggingface 🤗 Rustacean 🦀 ● BJJ competitor ● Lifelong Martial Artist

加入 March 2013

444 正在關注 643 粉絲

Dirhousssi Amine@DirhousssiAmine

2026.09.14 11:45

We ran async GRPO across HF Jobs with the trainer and vLLM on separate machines, no NCCL, and no shared disk. The adapter is mounted at the same path in every Job. LoRA training has landed in TRL's AsyncGRPOTrainer 🎉🎉🎉 The trainer now syncs only the adapter to vLLM ~few MB for a 1.5B model. That changes what the setup can look like. We ran async GRPO across HF Jobs with the trainer and vLLM on separate machines, no NCCL, no shared disk. The adapter travels through a 🪣 mounted on the same path in every Job. Blogpost 🧵