Spaces:
Sleeping
Sleeping
Commit History
-
feat: Add batch conversion scripts for CAM++ models
- Implemented `batch_convert.py` for programmatic conversion of multiple CAM++ models without Gradio UI. - Created `batch_convert.sh` for shell-based batch conversion. - Introduced `convert_cli.py` for command line interface model conversion with quantization options. - Added `inspect_model.py` to inspect the structure of ModelScope CAM++ models. - Developed `mlx_campp_modelscope.py` for a clean implementation of CAM++ model architecture in MLX. - Enhanced logging and error handling across scripts for better user experience.656e7f6 -
Add comprehensive status checking: only allow upload if conversion is perfect with 100% verification rate and no errors
17bd838 -
Fix critical conversion issues: remove incorrect Conv1d transpose, fix parameter mappings, disable fake parameter generation, update verification logic, improve test
77877c8 -
Refactor activation script and enhance conversion utilities; add parameter mapping and filtering logic; update requirements and add test for parameter mapping
ca0ebee
-
Add CAM++ model implementation and conversion utilities
- Implemented the CAM++ model architecture in mlx_campp.py, including dense layers, context-aware masking, and multi-granularity pooling. - Created conversion utilities in conversion_utils.py to convert PyTorch model weights to MLX format, including quantization and verification methods. - Added a pre-deployment validation script to validate project structure and content before deployment. - Included example usage and testing functions for model creation and conversion.368806f