Issue after updating torch

[Bearmancer]

How should I get rid of this error?

C:\Users\Lance\AppData\Local\Programs\Python\Python311\Lib\site-packages\whisper\__init__.py:146: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  checkpoint = torch.load(fp, map_location=device)```

Torch v: 2.4.1+cpu

[DJH02]

Actually, this is not an error, it is just a warning. If you set weights_only=True, this warning will be solved.

[Bearmancer]

This is what ChatGPT told me:

The warning you are seeing indicates that in the future, the default behavior of torch.load will change to set the parameter weights_only=True, which enhances security by limiting the types of objects that can be loaded. To address this warning now, you can explicitly set weights_only=True when using torch.load. If you are working with a library or package like Whisper, you may need to modify the code or check if there is an update to the package that already accounts for this.

Here is how you can modify the relevant part of the code:

Locate the line in the __init__.py file where torch.load is used.
Update the code by explicitly setting the weights_only=True argument when calling torch.load.
# Original code in __init__.py
model = torch.load('model_file.pt')

# Modify it to:
model = torch.load('model_file.pt', weights_only=True)

Upon searching the file I could only file this:

with (
        io.BytesIO(checkpoint_file) if in_memory else open(checkpoint_file, "rb")
    ) as fp:
        checkpoint = torch.load(fp, map_location=device)
    del checkpoint_file

I couldn't find what GPT suggested.

[Bearmancer]

And how does one do that?

…

________________________________ From: DJH02 ***@***.***> Sent: 13 September 2024 08:23 To: openai/whisper ***@***.***> Cc: Bearmancer ***@***.***>; Author ***@***.***> Subject: Re: [openai/whisper] Issue after updating torch (Discussion #2330) Actually, this is not an error, it is just a warning. If you set weights_only=True, this warning will be solved. — Reply to this email directly, view it on GitHub<#2330 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AQWHKKSW4IF7UFBMQ6VVHG3ZWJHSBAVCNFSM6AAAAABN4Z3RSCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTANRTGI2TONA>. You are receiving this because you authored the thread.

[mikenoname-github]

How does one 'set weights_only=True' when using Colab which is installing Whisper from the cloud each time? Can we download Whisper, fix it and store it in our own repository somehow?

[mikenoname-github]

Bump