[Feature] Added logging and evaluation functionality

[Saf9933]

Summary
The two key features:

• Added logging to track model loading, configuration, task execution, and evaluation.
• Added an evaluation function that calculates accuracy after task execution using a test dataset loaded from JSON.
  README Updates:
• Updated the README.md file to include a Contributing section.
  The section explains how others can fork the repository, create new branches, and submit Pull Requests to contribute to the project.

Device Compatibility

• The code is written to work on systems with or without CUDA support. It will automatically detect if CUDA is available and use it; otherwise, it defaults to the CPU.
  Testing issue for my MacBook
  Due to hardware limitations (MacBook M1, no CUDA support), I was unable to fully test the code. However, I structured it to run efficiently on a system with CUDA-enabled hardware. I request that someone with the necessary hardware verify the functionality thank you :).

Files Changed:

• mlora_train.py: Added logging, dataset loading, and evaluation.
• README.md: Contributing section

[yezhengmao1]

it's better to create a new file and new class .

[Saf9933]

Thank you for the feedback! I’ve now refactored the dataset loading and evaluation into a separate module (dataset_utils.py) as suggested you can check it out!

[yezhengmao1]

can we split the function (dataset loading and evaluation) to another file or module?

[yezhengmao1]

can we move this file into a new directory "dataset" ?