Commit
•
2a7d0aa
1
Parent(s):
61df9ad
configs
Browse files
config.json
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "MochiTransformer3DModel",
|
| 3 |
+
"_diffusers_version": "0.32.0.dev0",
|
| 4 |
+
"_name_or_path": "mochi-1-diffusers",
|
| 5 |
+
"activation_fn": "swiglu",
|
| 6 |
+
"attention_head_dim": 128,
|
| 7 |
+
"in_channels": 12,
|
| 8 |
+
"max_sequence_length": 256,
|
| 9 |
+
"num_attention_heads": 24,
|
| 10 |
+
"num_layers": 48,
|
| 11 |
+
"out_channels": null,
|
| 12 |
+
"patch_size": 2,
|
| 13 |
+
"pooled_projection_dim": 1536,
|
| 14 |
+
"qk_norm": "rms_norm",
|
| 15 |
+
"quantization_config": {
|
| 16 |
+
"_load_in_4bit": true,
|
| 17 |
+
"_load_in_8bit": false,
|
| 18 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 19 |
+
"bnb_4bit_quant_storage": "uint8",
|
| 20 |
+
"bnb_4bit_quant_type": "nf4",
|
| 21 |
+
"bnb_4bit_use_double_quant": false,
|
| 22 |
+
"llm_int8_enable_fp32_cpu_offload": false,
|
| 23 |
+
"llm_int8_has_fp16_weight": false,
|
| 24 |
+
"llm_int8_skip_modules": [
|
| 25 |
+
"final_layer",
|
| 26 |
+
"x_embedder.proj",
|
| 27 |
+
"t_embedder",
|
| 28 |
+
"pos_frequencies",
|
| 29 |
+
"t5",
|
| 30 |
+
"blocks.0",
|
| 31 |
+
"blocks.1",
|
| 32 |
+
"blocks.2",
|
| 33 |
+
"blocks.3",
|
| 34 |
+
"blocks.4",
|
| 35 |
+
"blocks.5",
|
| 36 |
+
"blocks.6",
|
| 37 |
+
"blocks.7",
|
| 38 |
+
"blocks.8",
|
| 39 |
+
"blocks.9",
|
| 40 |
+
"blocks.10",
|
| 41 |
+
"blocks.11",
|
| 42 |
+
"blocks.12",
|
| 43 |
+
"blocks.13",
|
| 44 |
+
"blocks.14",
|
| 45 |
+
"blocks.15",
|
| 46 |
+
"blocks.16",
|
| 47 |
+
"blocks.17",
|
| 48 |
+
"blocks.18",
|
| 49 |
+
"blocks.30",
|
| 50 |
+
"blocks.31",
|
| 51 |
+
"blocks.32",
|
| 52 |
+
"blocks.33",
|
| 53 |
+
"blocks.34",
|
| 54 |
+
"blocks.35",
|
| 55 |
+
"blocks.36",
|
| 56 |
+
"blocks.37",
|
| 57 |
+
"blocks.38",
|
| 58 |
+
"blocks.39",
|
| 59 |
+
"blocks.40",
|
| 60 |
+
"blocks.41",
|
| 61 |
+
"blocks.42",
|
| 62 |
+
"blocks.43",
|
| 63 |
+
"blocks.44",
|
| 64 |
+
"blocks.45",
|
| 65 |
+
"blocks.46",
|
| 66 |
+
"blocks.47"
|
| 67 |
+
],
|
| 68 |
+
"llm_int8_threshold": 6.0,
|
| 69 |
+
"load_in_4bit": true,
|
| 70 |
+
"load_in_8bit": false,
|
| 71 |
+
"quant_method": "bitsandbytes"
|
| 72 |
+
},
|
| 73 |
+
"text_embed_dim": 4096,
|
| 74 |
+
"time_embed_dim": 256
|
| 75 |
+
}
|
diffusion_pytorch_model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|