Image Segmentation
MLX
apple-silicon
video-matting
green-screen
blue-screen
chroma-key
vfx
fp32
fp16
quantized
Instructions to use cdelta/CorridorKey-v2-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use cdelta/CorridorKey-v2-MLX with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download cdelta/CorridorKey-v2-MLX --local-dir CorridorKey-v2-MLX
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
File size: 585 Bytes
a405f8d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 | {
"upstream": "nikopueringer/CorridorKey2.0",
"revision": "cf44a8db3f024bb49fd59daa408848ccf05f7477",
"tensor_count": 738,
"missing_or_extra": [],
"unequal_tensors": [],
"model_config_equal": true,
"public_safetensors_sha256": "c43596de98cc8866d577c8a9f9ea239acdf3f7a130ff24e25db4375a50590436",
"stats_keys": [
"cradio_mean",
"cradio_std",
"moge_mean",
"moge_std",
"rvm_mean",
"rvm_std"
],
"stats_equal": true,
"passed": true,
"method": "SHA-256 of conversion-input and official-release tensor bytes, with exact shape/dtype agreement"
}
|