This commit addresses compatibility issues with PyTorch 2.6+ and 2.7+ that cause runtime errors in MMEngine.

PyTorch 2.6+ JIT Compilation Fix:

• Add safe import mechanism for ZeroRedundancyOptimizer in zero_optimizer.py
• Temporarily disable JIT compilation during distributed optimizer import
• Apply fix for PyTorch >=2.6.0 where JIT compilation issues were introduced
• Graceful fallback when distributed optimizers are unavailable
• Resolves: RuntimeError during torch.distributed.optim import

PyTorch 2.6+ torch.load weights_only Fix:

• Add _safe_torch_load function in checkpoint.py with automatic version detection
• Handle weights_only parameter changes with safe globals for numpy arrays
• Fallback to weights_only=False for compatibility with existing checkpoints
• Resolves: "Weights only load failed" errors when loading models

Key Features:

• Maintains full backward compatibility with older PyTorch versions
• Automatic version detection and appropriate handling
• Conservative approach: only applies fixes to versions that need them
• Comprehensive error handling and user warnings
• Follows MMEngine coding standards

Version Support:

• PyTorch 2.6+ JIT compilation issues handled
• PyTorch 2.6+ weights_only parameter changes handled
• Full compatibility maintained for PyTorch 1.6-2.5

Files Changed:

• mmengine/optim/optimizer/zero_optimizer.py: Safe distributed optimizer import
• mmengine/runner/checkpoint.py: Safe torch.load with weights_only handling

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers. By the way, if you're not familiar with how to use pre-commit to fix lint issues or add unit tests, please refer to Contributing to OpenMMLab.

Motivation

Please describe the motivation of this PR and the goal you want to achieve through this PR.

Modification

Please briefly describe what modification is made in this PR.

BC-breaking (Optional)

Does the modification introduce changes that break the backward-compatibility of the downstream repos?
If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here, and update the documentation.

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
3. If the modification has potential influence on downstream projects, this PR should be tested with downstream projects, like MMDetection or MMPretrain.
4. The documentation has been modified accordingly, like docstring or example tutorials.

NOTE:this pr is generated by claude code