Hugging Face's logo
Activity Feed

AI & ML interests

RL Environments at Scale

Recent Activity
AdithyaSK  updated a Space about 22 hours ago
FineEnvs/fleurs-asr-env
AdithyaSK  published a Space about 22 hours ago
FineEnvs/fleurs-asr-env
AdithyaSK  updated a bucket about 22 hours ago
FineEnvs/fleurs-bucket
View all activity
AdithyaSK 
updated a Space about 22 hours ago
AdithyaSK 
published a Space about 22 hours ago
AdithyaSK 
updated 2 buckets about 22 hours ago
AdithyaSK 
published a bucket about 22 hours ago
AdithyaSK 
updated a bucket 1 day ago
AdithyaSK 
published a bucket 1 day ago
AdithyaSK 
updated a Space 1 day ago
sergiopaniego 
posted an update 1 day ago
view post
Post
110
ICYMI, Async GRPO in TRL now supports LoRA and we wrote a looong blog testing it

> the adapter is a few megabytes, so the weight sync is a file instead of an NCCL transfer
> 3 HF Jobs: 1 trainer and 2 vLLM replicas
> the adapter travels through an HF Storage Bucket mounted in all 3 at the same path
> a proxy in front of the replicas routes each rollout to the one already holding its KV prefix

https://huggingface.co/blog/asyncgrpo-lora-hfjobs
AdithyaSK 
updated 11 datasets 1 day ago