version and compatibility fix

[zakura61]

This PR updates the project to resolve issues with deprecated dependencies and ensure compatibility with current PyTorch versions.

Key changes include:

• Updated requirements.txt: Modernized dependencies for a clean installation via pip install -r requirements.txt. Verified successful installation without errors.
• Code modifications for PyTorch compatibility: Addressed deprecated code within the PyTorch sections. Verified that the commands in the README.md now execute correctly.
• Modifications for TensorFlow (Untested): This PR also includes updates to TensorFlow-related code, although this functionality remains untested. These changes address deprecations and may provide a basis for future TensorFlow compatibility, but further work might be required.

[Huah-Yin]

hi,scipy==1.14.1 this file is not exits ,i can't pip this one

it keeps failing, and pip tries to reconnect many times — but it can never succeed, because this version was never released

[zakura61]

Hi @Huah-Yin,

Actually, scipy==1.14.1 is a valid version and was released on PyPI: https://pypi.org/project/scipy/1.14.1/

As I mentioned in the PR description, I tested the installation using pip install -r requirements.txt successfully on Ubuntu before submitting.

However, if 1.14.1 consistently fails for you, could you try scipy==1.15.0? It's a newer stable version and should work fine as well.