Force weights_only=False in torch.load patch

Add diagnostic endpoint to debug torch.load issue

Add omegaconf safe globals for PyTorch 2.6+

Fix: remove duplicate torch import that undid patch

Add torch.load patch for weights_only default

Remove torch constraint - test if build works

Remove gradio version constraint - use HF default

Fix torch==2.5.1, clean up cache files

Clean up: remove torch.load patch (using torch<2.6), remove test endpoint

Pin torch<2.6 to avoid weights_only issue

Move torch.load patch to very first import

Monkey-patch torch.load weights_only=False for legacy checkpoints

Fix PyTorch 2.6 weights_only issue with omegaconf

Test model loading

Add diagnostic imports to test function

Add test endpoint for debugging

Remove spaces.GPU decorator for dedicated GPU

Pin pydantic<2.11 to fix schema bug

Upgrade to Gradio 6.x to fix pydantic schema bug

Switch to two Image components for simpler API

Add debug logging to reconstruction function

Fix File component examples crash - remove ios_mode

Switch to File component to fix API schema bug

Add explicit api_name for predict endpoint

Fix Gradio 5.x Gallery input format handling

Upgrade to Gradio 5.x for HfFolder fix

Pin huggingface-hub<1.0.0 to fix HfFolder import

Force rebuild

Update gradio to 4.44.0

Fix dependency versions

Enable ZeroGPU

Initial commit