Commit History

fix: Remove torch.cuda.is_available() to prevent CUDA init in main process
f9d4d6e

silveroxides commited on

fix: Pin convert_to_quant to commit hash to force reinstall
3099eb9

silveroxides commited on

fix: Use force_cpu=True to bypass CUDA checks on ZeroGPU
681221a

silveroxides commited on

fix: Move GPU decorator to completion-only stub to avoid timeout
d10c339

silveroxides commited on

temp: Disable MXFP8/NVFP4, remove comfy-kitchen (build failure)
aa41047

silveroxides commited on

deps: Add comfy-kitchen for MXFP8/NVFP4 quantization support
96f8e42

silveroxides commited on

fix: Lower GPU duration to 120s (ZeroGPU max)
16160d8

silveroxides commited on

fix: Use GPU for MXFP8/NVFP4 formats, CPU for others
28030ea

silveroxides commited on

fix: Move GPU decorator to minimal stub function, call at end of quantization
aee2946

silveroxides commited on

fix: Add @spaces.GPU with extended duration for ZeroGPU compatibility
c22291c

silveroxides commited on

perf: Run quantization on CPU, remove ZeroGPU timeout issue
4127859

silveroxides commited on

fix: Gradio 6.0 compatibility - move theme to launch(), remove deprecated show_api
aaee40e

silveroxides commited on

Update README.md
ea83b7f
verified

silveroxides commited on

Update README.md
13731f4
verified

silveroxides commited on

refactor: Replace subprocess-based CLI with programmatic API
c2c886b

silveroxides commited on