dependency compatibility
Hello I a struggling to use the model locally, can someone give me a compatible set of library versions or point me towards where I can get them?
for example
torch==x.x.x
transformers=x.x.x
pillow==x.x.x
Currently I have:
torch==2.5.1
transformers==4.46.2
pillow==11.0.0
I am getting this error:
This is part of the error:
.../lib/python3.11/site-packages/transformers/tokenization_utils_base.py:3527, in PreTrainedTokenizerBase.pad(self, encoded_inputs, padding, max_length, pad_to_multiple_of, padding_side, return_attention_mask, return_tensors, verbose)
3525 for i in range(batch_size):
3526 inputs = {k: v[i] for k, v in encoded_inputs.items()}
-> 3527 outputs = self._pad(
3528 inputs,
3529 max_length=max_length,
3530 padding_strategy=padding_strategy,
3531 pad_to_multiple_of=pad_to_multiple_of,
3532 padding_side=padding_side,
3533 return_attention_mask=return_attention_mask,
3534 )
3536 for key, value in outputs.items():
3537 if key not in batch_outputs:
TypeError: ChatGLM4Tokenizer._pad() got an unexpected keyword argument 'padding_side'
you can see it in requirements.txt in 3ML bot files. I suggest you to test this repo "https://github.com/nikravan1/3ML" on google colab.
bitsandbytes
git+https://github.com/huggingface/accelerate.git
Pillow
torch
torchvision
transformers==4.41.2
Numpy