| { | |
| "_class_name": "MochiTransformer3DModel", | |
| "_diffusers_version": "0.32.0.dev0", | |
| "_name_or_path": "mochi-1-diffusers", | |
| "activation_fn": "swiglu", | |
| "attention_head_dim": 128, | |
| "in_channels": 12, | |
| "max_sequence_length": 256, | |
| "num_attention_heads": 24, | |
| "num_layers": 48, | |
| "out_channels": null, | |
| "patch_size": 2, | |
| "pooled_projection_dim": 1536, | |
| "qk_norm": "rms_norm", | |
| "quantization_config": { | |
| "_load_in_4bit": true, | |
| "_load_in_8bit": false, | |
| "bnb_4bit_compute_dtype": "bfloat16", | |
| "bnb_4bit_quant_storage": "uint8", | |
| "bnb_4bit_quant_type": "nf4", | |
| "bnb_4bit_use_double_quant": false, | |
| "llm_int8_enable_fp32_cpu_offload": false, | |
| "llm_int8_has_fp16_weight": false, | |
| "llm_int8_skip_modules": [ | |
| "final_layer", | |
| "x_embedder.proj", | |
| "t_embedder", | |
| "pos_frequencies", | |
| "t5", | |
| "blocks.0", | |
| "blocks.1", | |
| "blocks.2", | |
| "blocks.3", | |
| "blocks.4", | |
| "blocks.5", | |
| "blocks.6", | |
| "blocks.7", | |
| "blocks.8", | |
| "blocks.9", | |
| "blocks.10", | |
| "blocks.11", | |
| "blocks.12", | |
| "blocks.13", | |
| "blocks.14", | |
| "blocks.15", | |
| "blocks.16", | |
| "blocks.17", | |
| "blocks.18", | |
| "blocks.30", | |
| "blocks.31", | |
| "blocks.32", | |
| "blocks.33", | |
| "blocks.34", | |
| "blocks.35", | |
| "blocks.36", | |
| "blocks.37", | |
| "blocks.38", | |
| "blocks.39", | |
| "blocks.40", | |
| "blocks.41", | |
| "blocks.42", | |
| "blocks.43", | |
| "blocks.44", | |
| "blocks.45", | |
| "blocks.46", | |
| "blocks.47" | |
| ], | |
| "llm_int8_threshold": 6.0, | |
| "load_in_4bit": true, | |
| "load_in_8bit": false, | |
| "quant_method": "bitsandbytes" | |
| }, | |
| "text_embed_dim": 4096, | |
| "time_embed_dim": 256 | |
| } | |