Fix dtype issues and update dependencies

[nick8592]

This MR includes three main updates to improve type compatibility and prevent runtime errors:

1. Update datasets.py
   Replace deprecated np.int with np.int32 for explicit dtype specification.

2. Update loss.py
   Fix RuntimeError caused by casting issues: ensure result type Float is properly handled instead of trying to cast to long int.

3. Update requirements.txt
   Modify dependencies to prevent the TypeError: No loop matching the specified signature and casting was found for ufunc greater error.