kotoba-tech

/

kotoba-whisper-v2.2

like 127

Follow

Kotoba Technologies 190

Not sure if this is still being updated but just in case it is...

Following the instructions in the model card results in errors. After a little searching it seems these maybe related to the transformers version used.

Transformers version I have installed is 5.2.0.

Suspect changing the transformers version in the model card install details to a specific working version would resolve.

Happy to provide more environment details if required.

Warnings / Errors:

1. torch_dtype is deprecated! Use dtype instead!

Easy fix to script on model card ...
torch_dtype=torch_dtype,
to:
dtype=torch_dtype,

2. AttributeError: 'SegmentationModel' object has no attribute 'all_tied_weights_keys'.

Full error message:
Loading weights: 100%|█| 54/54 [00:00<00:00, 2057.45it/s, Materializing param=model.sincnet.wav_norm1d.we
Traceback (most recent call last):
File "/home/fast-whisper-user/transcription/test_program.py", line 12, in
pipe = pipeline(
^^^^^^^^^
File "/home/fast-whisper-user/transcription/transcription_env/lib64/python3.12/site-packages/transformers/pipelines/init.py", line 1028, in pipeline
return pipeline_class(model=model, task=task, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/fast-whisper-user/.cache/huggingface/modules/transformers_modules/kotoba_hyphen_tech/kotoba_hyphen_whisper_hyphen_v2_dot_2/9d33482a0eb9b57f1ad80708e8ac5538246d8355/kotoba_whisper.py", line 88, in init
self.model_speaker_diarization = SpeakerDiarization(
^^^^^^^^^^^^^^^^^^^
File "/home/fast-whisper-user/.cache/huggingface/modules/transformers_modules/kotoba_hyphen_tech/kotoba_hyphen_whisper_hyphen_v2_dot_2/9d33482a0eb9b57f1ad80708e8ac5538246d8355/kotoba_whisper.py", line 45, in init
self.pipeline._segmentation.model = SegmentationModel().from_pretrained(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/fast-whisper-user/transcription/transcription_env/lib64/python3.12/site-packages/transformers/modeling_utils.py", line 4073, in from_pretrained
loading_info = cls._finalize_model_loading(model, load_config, loading_info)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/fast-whisper-user/transcription/transcription_env/lib64/python3.12/site-packages/transformers/modeling_utils.py", line 4214, in _finalize_model_loading
model._adjust_tied_keys_with_tied_pointers(loading_info.missing_and_mismatched())
File "/home/fast-whisper-user/transcription/transcription_env/lib64/python3.12/site-packages/transformers/modeling_utils.py", line 4458, in _adjust_tied_keys_with_tied_pointers
self.all_tied_weights_keys.update(tied_weights_keys_by_pointers)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/fast-whisper-user/transcription/transcription_env/lib64/python3.12/site-packages/torch/nn/modules/module.py", line 1965, in getattr
raise AttributeError(
AttributeError: 'SegmentationModel' object has no attribute 'all_tied_weights_keys'. Did you mean: '_tied_weights_keys'?