OSError: We couldn't connect to 'https://huggingface.co' to load this file, couldn't find it in the cached files and it looks like facebook/mms-1b-all is not the path to a directory containing a file named config.json.
Checkout your internet connection or see how to run the library in offline mode at 'https://huggingface.co/docs/transformers/installation#offline-mode'.
Failed to load pretrained models. Any solutions?
Ok now I firstly download it via colab and then to my own env. Another problem is Traceback (most recent call last):
File "asr_infer_in_a_row.py", line 17, in
model = Wav2Vec2ForCTC.from_pretrained(model_path)
File "/home/zhangzheyu/miniconda3/envs/mms/lib/python3.8/site-packages/transformers/modeling_utils.py", line 2446, in from_pretrained
OSError: Error no file named pytorch_model.bin, tf_model.h5, model.ckpt.index or flax_model.msgpack found in directory /home/zhangzheyu/fairseq/examples/mms/mms-1b-all.
It seems that a pytorch_model.bin file is needed while the repo does not have the file.
The model path should be set to the model id on the Hub (
facebook/mms-1b-all) or a local folder where you've cloned the repo
Here is an example on how one can use MMS models - https://huggingface.co/spaces/mms-meta/MMS/blob/main/asr.py#L37
Yes, i did connect to the internet. However, you know, In china you cannot directly use from_pretrain as the huggingface is blocked. So you have to manually download the models via git lfs. The pytorch_model.bin is missing somehow by the git clone. I tried to replace the file from facebook/mms-1b but failed. Do you know the right way to fix it? @sanchit-gandhi