Unable to load checkpoint about golden-gaze HOT 3 OPEN

@heinrichreimer So sorry for this late reply. I don't know why I didn't receive an email notification of this issue. Are you using the exact versions of dependent libraries mentioned at: https://github.com/UWaterlooIR/golden-gaze#environment?

from golden-gaze.

No, some of my dependencies have higher versions, but nothing that shouldn't be compatible:

beautifulsoup4==4.11.1
nltk==3.8.1
pytorch-lightning==1.9.0
scikit-learn==1.2.1
torch==1.12.1
transformers==4.26.0

As the error message mentioned some issue regarding the ZIP file, I tried to unzip the file, which should normally work for PyTorch lightning files:

Archive:  model.ckpt
  End-of-central-directory signature not found.  Either this file is not
  a zipfile, or it constitutes one disk of a multi-part archive.  In the
  latter case the central directory and zipfile comment will be found on
  the last disk(s) of this archive.
unzip:  cannot find zipfile directory in one of model.ckpt or
        model.ckpt.zip, and cannot find model.ckpt.ZIP, period.

Is the checkpoint maybe part of a multi-file checkpoint?
The checkpoint size also seems to be rather small (1.5GB) whereas the base model is 3GB.

from golden-gaze.

It's not a zipped file. Could you post your code where you load the checkpoint?
My code at lines 184-186 in this file: (https://github.com/UWaterlooIR/golden-gaze/blob/main/stance_detection/main.py) works with the checkpoint file (https://github.com/UWaterlooIR/golden-gaze/releases/download/model/model.ckpt).

model = models.T5StanceDetector.load_from_checkpoint(f'output/{params["checkpoint_path"]}',
                                                       params=params, tokenizer=inference_set.tokenizer,
                                                       data=inference_data)

from golden-gaze.