Emotional Model Torch Error

[kleywinn]

Hi I'm having a hard time running yuna-ai. This is the log I'm getting when running python3 index.py:

It seems like torch is in the wrong version (from what I checked it's 2.2.2. And I'm using python 3.8.

Using device: mps
Traceback (most recent call last):
  File "index.py", line 208, in <module>
    yuna_server = YunaServer()
  File "index.py", line 35, in __init__
    self.chat_generator = ChatGenerator(self.config)
  File "/Users/kleywinn/Documents/yuna-ai/lib/generate.py", line 19, in __init__
    self.classifier = pipeline("text-classification", model=f"{config['server']['agi_model_dir']}yuna-emotion")
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/pipelines/__init__.py", line 905, in pipeline
    framework, model = infer_framework_load_model(
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/pipelines/base.py", line 292, in infer_framework_load_model
    raise ValueError(
ValueError: Could not load model lib/models/agi/yuna-emotion with any of the following classes: (<class 'transformers.models.auto.modeling_auto.AutoModelForSequenceClassification'>, <class 'transformers.models.roberta.modeling_roberta.RobertaForSequenceClassification'>). See the original errors:

while loading with AutoModelForSequenceClassification, an error is thrown:
Traceback (most recent call last):
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/modeling_utils.py", line 532, in load_state_dict
    return torch.load(
  File "/opt/homebrew/lib/python3.8/site-packages/torch/serialization.py", line 1005, in load
    with _open_zipfile_reader(opened_file) as opened_zipfile:
  File "/opt/homebrew/lib/python3.8/site-packages/torch/serialization.py", line 457, in __init__
    super().__init__(torch._C.PyTorchFileReader(name_or_buffer))
RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/modeling_utils.py", line 541, in load_state_dict
    if f.read(7) == "version":
  File "/opt/homebrew/Cellar/python@3.8/3.8.19/Frameworks/Python.framework/Versions/3.8/lib/python3.8/codecs.py", line 322, in decode
    (result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 64: invalid start byte

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/pipelines/base.py", line 279, in infer_framework_load_model
    model = model_class.from_pretrained(model, **kwargs)
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/models/auto/auto_factory.py", line 563, in from_pretrained
    return model_class.from_pretrained(
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/modeling_utils.py", line 3335, in from_pretrained
    state_dict = load_state_dict(resolved_archive_file)
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/modeling_utils.py", line 553, in load_state_dict
    raise OSError(
OSError: Unable to load weights from pytorch checkpoint file for 'lib/models/agi/yuna-emotion/pytorch_model.bin' at 'lib/models/agi/yuna-emotion/pytorch_model.bin'. If you tried to load a PyTorch model from a TF 2.0 checkpoint, please set from_tf=True.

while loading with RobertaForSequenceClassification, an error is thrown:
Traceback (most recent call last):
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/modeling_utils.py", line 532, in load_state_dict
    return torch.load(
  File "/opt/homebrew/lib/python3.8/site-packages/torch/serialization.py", line 1005, in load
    with _open_zipfile_reader(opened_file) as opened_zipfile:
  File "/opt/homebrew/lib/python3.8/site-packages/torch/serialization.py", line 457, in __init__
    super().__init__(torch._C.PyTorchFileReader(name_or_buffer))
RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/modeling_utils.py", line 541, in load_state_dict
    if f.read(7) == "version":
  File "/opt/homebrew/Cellar/python@3.8/3.8.19/Frameworks/Python.framework/Versions/3.8/lib/python3.8/codecs.py", line 322, in decode
    (result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 64: invalid start byte

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/pipelines/base.py", line 279, in infer_framework_load_model
    model = model_class.from_pretrained(model, **kwargs)
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/modeling_utils.py", line 3335, in from_pretrained
    state_dict = load_state_dict(resolved_archive_file)
  File "/opt/homebrew/lib/python3.8/site-packages/transformers/modeling_utils.py", line 553, in load_state_dict
    raise OSError(
OSError: Unable to load weights from pytorch checkpoint file for 'lib/models/agi/yuna-emotion/pytorch_model.bin' at 'lib/models/agi/yuna-emotion/pytorch_model.bin'. If you tried to load a PyTorch model from a TF 2.0 checkpoint, please set from_tf=True.

[yukiarimo]

1. How did you install the torch?
2. Are you using conda?
3. What is your machine and OS?
4. Have you installed all of the requirements (especially transformers)?
5. As a hotfix, in the lib/generate.py

replace this line: self.classifier = pipeline("text-classification", model=f"{config['server']['agi_model_dir']}yuna-emotion")

With just: self.classifier = ""

[kleywinn]

1. I used pip3 install torch
2. No, I installed python via brew as a matter of fact I'm using the commands python3.8 and pip3.8 to specify the version.
3. MacOS M1 Sonoma
4. I believe so, at least everything from the installation suite.
5. I did and it works, but I'm getting an error in the messaging response due to the lack of classifier

   File "/Users/kleywinn/Documents/yuna-ai/lib/generate.py", line 77, in generate
   response_add = self.classifier(response)[0]['label']
   TypeError: 'str' object is not callable

[yukiarimo]

1. The official command is pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cpu (better to use that for more features)
2. Even if it works for you, I'll highly recommend installing Anaconda, creating a virtual (condo's) environment, and then doing your stuff!
3. Glad to hear that!
4. Yeah, I see.
5. So, why not to just disable emotions in the static/config.json?!

[kleywinn]

Ok, I'll give it a try with Anaconda. I disabled emotions and it worked indeed! I'm closing this issue for now. Thanks for the assistance.