huggingface / huggingface/trl

grpo init_communicator ipv6 can not reach

Open
#3,164 4 comments 1 reaction 0 assignees View on GitHub
✨ enhancement 🏋 GRPO
Dominant language
Python
Stars
19.3k
Forks
3k
Avg merge
1d 20h
Merged PRs (30d)
194

Description

Hello, when I use GRPO provided by trl for training, I found that when only IPv6 is available, there is the following code in `trl/extras/vllm_client.py` and `trl/scripts/vllm_serve.py`:

```python
pg = StatelessProcessGroup.create(host=self.host, port=self.group_port, rank=self.rank, world_size=world_size)
```

There is a timeout issue, it seems that communication cannot be established through IPv6. When I change the above code to the following, it seems to work, but I am not sure if it is correct:

```python
temp_host = self.host.replace("[", "").replace("]", "")
pg = StatelessProcessGroup.create(host=temp_host, port=self.group_port, rank=self.rank, world_size=world_size)
```

Have you tested it in an IPv6 environment?

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.