Spaces:
Running
on
Zero
Running
on
Zero
| # Gradio (REQUIRED - specific version) | |
| gradio==5.49.1 | |
| # ZeroGPU support | |
| spaces==0.30.4 | |
| # PyTorch (pinned for FlashAttention compatibility) | |
| torch==2.8.0 | |
| torchvision | |
| torchaudio | |
| # FlashAttention (prebuilt wheel for PyTorch 2.8, Python 3.10, CUDA 12) | |
| flash-attn @ https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.8cxx11abiFALSE-cp310-cp310-linux_x86_64.whl | |
| # Transformers and ML libraries | |
| transformers==4.56.2 | |
| accelerate>=0.28.0 | |
| timm==1.0.19 | |
| peft==0.15.2 | |
| # Vision and image processing | |
| opencv-python | |
| pillow>=9.4.0 | |
| pycocotools | |
| git+https://github.com/cocodataset/panopticapi.git | |
| git+https://github.com/facebookresearch/segment-anything.git | |
| # Data handling | |
| numpy==1.26.4 | |
| pydantic>=2.10.1 | |
| protobuf<=3.20.3 | |
| sentencepiece | |
| # Utilities | |
| httpx>=0.24.1,<1.0 | |
| requests | |
| inflect | |
| distinctipy | |
| attrdict | |
| # Video processing (optional) | |
| pytorchvideo | |
| # VLLM (optional - for evaluation) | |
| # vllm==0.7.3 | |
| # xTuner (for training - not needed for inference) | |
| # xtuner==0.2.0rc0 | |