Commit History
Fix image type: pass numpy array to inference instead of PIL Image 7be0ea0
Increase GPU duration to 600s for cold model load 09ba6b2
Fix mask type: pass numpy array to inference instead of PIL Image 65c471a
Remove pip install from Dockerfile — let HF handle requirements.txt via pyenv pip 3cc5bd4
Pre-clone DINOv2 into torch hub cache to avoid GitHub access at inference time 6a9de10
Update app.py 89682b4 verified
Fix config_path: use pipeline.yaml from downloaded checkpoints, not GitHub repo 46acce9
Add spconv-cu121; set LIDRA_SKIP_INIT to skip missing conda init module 0f03a92
Add Dockerfile for cached builds; fix MCP URL db6bdba
Add all missing sam-3d-objects dependencies 871be8a
Update requirements.txt 4f6e871 verified
Update requirements.txt 1757047 verified
Pin torch 2.5.1+cu121, add pytorch3d 0.7.8 and kaolin 0.17.0 wheels 56fd225
Add kaolin shim for torch 2.10+cu128 (no prebuilt wheel available) 1a197df
Dev Mode commited on