Instructions to use mlx-community/RelayVSR-bf16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/RelayVSR-bf16 with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download mlx-community/RelayVSR-bf16 --local-dir RelayVSR-bf16
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Download config.json from mlx-community/RelayVSR-bf16: direct link, hf CLI and curl.
- Browser
- Download file 2.55 kB
-
https://huggingface.co/mlx-community/RelayVSR-bf16/resolve/main/config.json
- Command line
-
hf download hf://mlx-community/RelayVSR-bf16/config.json
-
curl -L -o config.json https://huggingface.co/mlx-community/RelayVSR-bf16/resolve/main/config.json
2.55 kB
| { | |
| "model": "RelayVSR", | |
| "source": { | |
| "repo": "kopper/RelayVSR", | |
| "code": "kopperx/RelayVSR", | |
| "license": "Apache-2.0", | |
| "base_model": "Wan-AI/Wan2.2-TI2V-5B (Apache-2.0)", | |
| "wan_sha256": "638acb6a27c17ba87b2e66a1770f17e61a99ad87cfff6fa3d3e05c2d57f2eb98", | |
| "flash_sha256": "7b8b8c99275fd5c4501b0589cc7b4c4506456a5df83160ef4e207e147b30f714" | |
| }, | |
| "generator": { | |
| "config": { | |
| "patch_size": [ | |
| 1, | |
| 2, | |
| 2 | |
| ], | |
| "num_attention_heads": 24, | |
| "attention_head_dim": 128, | |
| "in_channels": 48, | |
| "out_channels": 48, | |
| "text_dim": 4096, | |
| "freq_dim": 256, | |
| "ffn_dim": 14336, | |
| "num_layers": 30, | |
| "cross_attn_norm": true, | |
| "qk_norm": "rms_norm_across_heads", | |
| "eps": 1e-06, | |
| "image_dim": null, | |
| "rope_max_seq_len": 1024, | |
| "pos_embed_seq_len": null, | |
| "lq_proj_hidden_dim1": 768, | |
| "lq_proj_hidden_dim2": 3072, | |
| "use_block_causal_attention": true, | |
| "num_frames_per_block": 1, | |
| "local_attn_size": -1, | |
| "sink_size": 0, | |
| "flex_attention_block_size": 128, | |
| "spatial_window_size": [ | |
| 22, | |
| 40 | |
| ], | |
| "spatial_attention_backend": "triton", | |
| "spatial_attention_mode": "local", | |
| "temporal_context_frames": 4, | |
| "temporal_rope": { | |
| "policy": "fixed_end", | |
| "training_frames": 8 | |
| } | |
| }, | |
| "training_step": "5000", | |
| "weight_variant": "ema", | |
| "parameters": 5031059136, | |
| "dtype": "bf16" | |
| }, | |
| "decoder": { | |
| "config": { | |
| "condition_channels": 48, | |
| "encoder_channels": [ | |
| 64, | |
| 128, | |
| 128 | |
| ], | |
| "gradient_checkpointing": "none", | |
| "mlp_ratio": 2.6666666666666665, | |
| "norm_eps": 1e-05, | |
| "rope_theta": 10000.0, | |
| "spatial_attention": { | |
| "backend": "auto", | |
| "boundary": "truncated", | |
| "condition_self_window_size": [ | |
| 48, | |
| 48 | |
| ], | |
| "end_condition_window_size": [ | |
| 48, | |
| 48 | |
| ], | |
| "interval": "half_open", | |
| "start_condition_window_size": [ | |
| 48, | |
| 48 | |
| ], | |
| "type": "query_centered", | |
| "video_window_size": [ | |
| 24, | |
| 24 | |
| ] | |
| }, | |
| "spatial_head_dim": 1024, | |
| "temporal_rope_policy": "current_relative", | |
| "training_execution": "streaming", | |
| "upscale_factor": 4, | |
| "variant": "S", | |
| "window_size": 2, | |
| "condition_mode": "endpoints" | |
| }, | |
| "training_step": "60000", | |
| "weight_variant": "ema", | |
| "parameters": 31934208, | |
| "dtype": "fp32" | |
| }, | |
| "prompt": { | |
| "shape": [ | |
| 1, | |
| 512, | |
| 4096 | |
| ], | |
| "dtype": "bf16" | |
| }, | |
| "sampling": { | |
| "num_train_timesteps": 1000, | |
| "shift": 5.0, | |
| "timesteps": [ | |
| 1000 | |
| ], | |
| "guidance_scale": 1.0 | |
| }, | |
| "layout": "keys verbatim; conv weights channels-last; lq_proj norm weights flattened" | |
| } |