| { | |
| "_class_name": "LongCatVideoAvatarTransformer3DModel", | |
| "architectures": [ | |
| "LongCatVideoAvatarTransformer3DModel" | |
| ], | |
| "_diffusers_version": "0.32.0", | |
| "in_channels": 16, | |
| "out_channels": 16, | |
| "hidden_size": 4096, | |
| "depth": 48, | |
| "num_heads": 32, | |
| "caption_channels": 4096, | |
| "model_max_length": 512, | |
| "mlp_ratio": 4, | |
| "adaln_tembed_dim": 512, | |
| "frequency_embedding_size": 256, | |
| "patch_size": [ | |
| 1, | |
| 2, | |
| 2 | |
| ], | |
| "enable_flashattn3": false, | |
| "enable_flashattn2": true, | |
| "enable_xformers": false, | |
| "enable_bsa": false, | |
| "bsa_params": null, | |
| "cp_split_hw": null, | |
| "text_tokens_zero_pad": true, | |
| "audio_window": 5, | |
| "audio_block": 5, | |
| "audio_channel": 1280, | |
| "intermediate_dim": 512, | |
| "output_dim": 768, | |
| "context_tokens": 32, | |
| "vae_scale": 4, | |
| "audio_prenorm": false, | |
| "class_range": 24, | |
| "class_interval": 4 | |
| } |