Support latest Transformers and new cache design

[mrwyattii]

[mayank31398]

from transformers.utils import WEIGHTS_NAME, WEIGHTS_INDEX_NAME, cached_path, hf_bucket_url ImportError: cannot import name 'cached_path' from 'transformers.utils fails with transformers 4.23.1 :(

path = snapshot_download(
    repo_id=model_name,
    allow_patterns=["*"],
    local_files_only=is_offline_mode(),
    cache_dir=os.getenv("TRANSFORMERS_CACHE", None)
)

How about something like this ^^?

[mrwyattii]

from transformers.utils import WEIGHTS_NAME, WEIGHTS_INDEX_NAME, cached_path, hf_bucket_url ImportError: cannot import name 'cached_path' from 'transformers.utils fails with transformers 4.23.1 :(

path = snapshot_download(
    repo_id=model_name,
    allow_patterns=["*"],
    local_files_only=is_offline_mode(),
    cache_dir=os.getenv("TRANSFORMERS_CACHE", None)
)

How about something like this ^^?

Thanks for the suggestion. The implementation I chose is similar to this. If you have a chance, try it out and let me know if you run into any bugs :)

[mayank31398]

I have tested this @mrwyattii and it works fine. One thing to note is that: earlier I had to pass the path as: TRANSFORMERS_CACHE/models-bigscience-bloom and now it is just: TRANSFORMERS_CACHE.

I would say the newer change you have made is much better and intuitive. Thanks for this PR. :)

[mayank31398]

I would say, after a few versions, we can drop support for older transformers maybe? I don't really think its needed since I think there is only a handful of people using MII for BLOOM :)

[mayank31398]

Can we merge this?

[mrwyattii]

Can we merge this?

yes, we'll get this merged very soon! sorry for the delay