Fix conda environment dependency conflicts causing CI build failures

[Copilot]

Problem

The conda environment file (environment.yml) contained package versions that don't exist or are incompatible, causing the CI build to fail when trying to resolve dependencies. This prevented the GitHub Actions workflow from successfully creating the conda environment.

Root Causes

1. PyTorch 2.8.0 - This version doesn't exist yet; the latest stable releases are 2.5.x and 2.6.x
2. TensorFlow 2.20.0 - This version hasn't been released; current stable versions are 2.16.x and 2.17.x
3. Deprecated cudatoolkit - This package is deprecated and causes conflicts with modern PyTorch installations
4. scikit-learn 1.7.2 - This version is not released; stable versions are 1.5.x and 1.6.x
5. torchvision 0.23.0 - This version doesn't exist and must be compatible with the PyTorch version

Changes

Updated all package versions to existing, compatible releases with version ranges to ensure stability:

Package	Before	After	Reason
PyTorch	>=2.8.0	>=2.5.0,<2.7.0	2.8.0 doesn't exist
TensorFlow	>=2.20.0	>=2.16.0,<2.18.0	2.20.0 is not available
CUDA Support	cudatoolkit	pytorch-cuda=12.1	cudatoolkit is deprecated
scikit-learn	>=1.7.2	>=1.5.0,<1.7.0	1.7.2 is not released
torchvision	>=0.23.0	>=0.20.0,<0.22.0	Must match PyTorch version

Files Modified

• environment.yml - Fixed conda package versions
• requirements.txt - Synchronized pip package versions
• README.md - Updated documentation to reflect correct versions

Benefits

• ✅ Conda can now successfully resolve all dependencies
• ✅ CI workflow will complete without version conflicts
• ✅ Version ranges provide stability while allowing minor updates
• ✅ Modern CUDA support with pytorch-cuda=12.1
• ✅ All packages are now compatible with each other

Testing

• Validated YAML syntax
• Passed flake8 linting checks
• Integration tests pass successfully
• Version compatibility verified

This fix ensures the project's dependencies can be installed reliably in CI/CD pipelines and by contributors.

Original prompt

This PR fixes the conda environment dependency conflicts that were causing the CI build to fail.

Issues Fixed:

• PyTorch version incompatibility: Updated from >=2.8.0 to >=2.5.0,<2.7.0
• TensorFlow version doesn't exist: Changed from >=2.20.0 to >=2.16.0,<2.18.0 (2.20.0 is not available)
• Deprecated cudatoolkit: Replaced with pytorch-cuda=12.1 for proper CUDA support
• scikit-learn version doesn't exist: Updated from >=1.7.2 to >=1.5.0,<1.7.0 (1.7.2 is not released)
• torchvision compatibility: Updated from >=0.23.0 to >=0.20.0,<0.22.0 to match PyTorch version

Changes Made:

• Updated all package versions to existing, compatible releases
• Used version ranges to ensure compatibility between packages
• Replaced deprecated cudatoolkit with pytorch-cuda=12.1

These changes should allow conda to successfully resolve all dependencies and fix your CI workflow.

---

💬 Share your feedback on Copilot coding agent for the chance to win a $200 gift card! Click here to start the survey.