"ImportError: Package `num2words` is required to run SmolVLM processor" getting this issue wwhile importing SmolVLM2 from AutoProcessor
I have installed the transformer using the following command "pip install git+https://github.com/huggingface/[email protected]".
and using the following code to import the model ,
processor = AutoProcessor.from_pretrained(model_path)
model = AutoModelForImageTextToText.from_pretrained(
model_path,
torch_dtype=torch.bfloat16,
_attn_implementation="flash_attention_2"
).to("cuda")```
it's showing the error mentioned in the title. Explicitly I have installed the 'num2words' package using pip, still same error is showing.
is there a particular version I need to install ? (PS: I have restarted the runtime multiple times, that didn't work.
I modify the "lib/python3.12/site-packages/transformers/models/smolvlm/processing_smolvlm.py" file.
The line 48 show this:
These conditional isn't executing as true... so import "from num2words import num2words" before this line and it will work.
I'm also facing the issue. Any idea how to move ahead?
When running in colab, be sure to restart your runtime after installing packages. That should fix your issue.
I have multiple times and its giving the same error :(
Can the package be updated to avoid requiring a runtime restart? I want to use this for educational purposes, and the workaround is forcing me to move on to alternatives, as the last thing I want learners to have to do is troubleshoot a basic example. If not, no worries, we will move on.