Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL
Hugging Face blog describes running async GRPO reinforcement learning with LoRA across HF Jobs using a storage bucket and proxy instead of NCCL.
1 stories, newest first · one card per story
Hugging Face blog describes running async GRPO reinforcement learning with LoRA across HF Jobs using a storage bucket and proxy instead of NCCL.
Today 1.6K readers · 2.5K views · 779 reads · all time 189K views