FSDP Example ReadTimeoutError

[nghtm]

 7: [rank80]: urllib3.exceptions.ReadTimeoutError: HTTPSConnectionPool(host='huggingface.co', port=443): Read timed out. (read timeout=10)

Running FSDP example, 16 p5 nodes. The example worked with 8 nodes

[nghtm]

Suspected issue with train_util.py.

Suggested adding to line 19:

datasets.config.DOWNLOADER_TIMEOUT = 1800

and to line 478 (within streaming dataloader function)

tokenizer = AutoTokenizer.from_pretrained(tokenizer, timeout=1800)

[shimomut]

This issue resolved by this commit.
559e018

[github-actions]

This issue is stale because it has been open for 30 days with no activity.