Cosmos
Diffusers
Safetensors
cosmos3_omni
nvidia
cosmos3
vllm
vllm-omni
sglang
sglang-diffusion
text, image, video, audio, and action generation
omnimodel
Instructions to use nvidia/Cosmos3-Super with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Cosmos
How to use nvidia/Cosmos3-Super with Cosmos:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Diffusers
How to use nvidia/Cosmos3-Super with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("nvidia/Cosmos3-Super", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
Commit ·
6bd6b49
1
Parent(s): fe77b66
FP8 checkpoint for Cosmos3-Super (#16)
Browse files- FP8 checkpoint for Cosmos3-Super (eae12e423ab4717787f78c72cd4e1e4efdf5e296)
Co-authored-by: Wojciech Kutak <wkutak@users.noreply.huggingface.co>
- hf_quant_config.json +185 -0
- model.safetensors.index.json +0 -0
- transformer/config.json +212 -1
- transformer/{diffusion_pytorch_model-00001-of-00027.safetensors → diffusion_pytorch_model-00001-of-00014.safetensors} +2 -2
- transformer/{diffusion_pytorch_model-00002-of-00027.safetensors → diffusion_pytorch_model-00002-of-00014.safetensors} +2 -2
- transformer/{diffusion_pytorch_model-00003-of-00027.safetensors → diffusion_pytorch_model-00003-of-00014.safetensors} +2 -2
- transformer/{diffusion_pytorch_model-00004-of-00027.safetensors → diffusion_pytorch_model-00004-of-00014.safetensors} +2 -2
- transformer/diffusion_pytorch_model-00005-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00005-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00006-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00006-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00007-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00007-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00008-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00008-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00009-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00009-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00010-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00010-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00011-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00011-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00012-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00012-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00013-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00013-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00014-of-00014.safetensors +3 -0
- transformer/diffusion_pytorch_model-00014-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00015-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00016-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00017-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00018-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00019-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00020-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00021-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00022-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00023-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00024-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00025-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00026-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model-00027-of-00027.safetensors +0 -3
- transformer/diffusion_pytorch_model.safetensors.index.json +0 -0
- transformer/modelopt_state.pth +3 -0
hf_quant_config.json
ADDED
|
@@ -0,0 +1,185 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"block_quantize": null,
|
| 3 |
+
"calib_cfg": {
|
| 4 |
+
"method": "max"
|
| 5 |
+
},
|
| 6 |
+
"channel_quantize": null,
|
| 7 |
+
"disable_conv_quantization": false,
|
| 8 |
+
"forward_loop": null,
|
| 9 |
+
"modelopt_config": {
|
| 10 |
+
"algorithm": "max",
|
| 11 |
+
"quant_cfg": [
|
| 12 |
+
{
|
| 13 |
+
"enable": false,
|
| 14 |
+
"quantizer_name": "*"
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"cfg": {
|
| 18 |
+
"axis": null,
|
| 19 |
+
"fake_quant": false,
|
| 20 |
+
"num_bits": [
|
| 21 |
+
4,
|
| 22 |
+
3
|
| 23 |
+
]
|
| 24 |
+
},
|
| 25 |
+
"quantizer_name": "*weight_quantizer"
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"cfg": {
|
| 29 |
+
"axis": null,
|
| 30 |
+
"num_bits": [
|
| 31 |
+
4,
|
| 32 |
+
3
|
| 33 |
+
]
|
| 34 |
+
},
|
| 35 |
+
"quantizer_name": "*input_quantizer"
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"enable": false,
|
| 39 |
+
"parent_class": "nn.BatchNorm1d",
|
| 40 |
+
"quantizer_name": "*"
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"enable": false,
|
| 44 |
+
"parent_class": "nn.BatchNorm2d",
|
| 45 |
+
"quantizer_name": "*"
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"enable": false,
|
| 49 |
+
"parent_class": "nn.BatchNorm3d",
|
| 50 |
+
"quantizer_name": "*"
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"enable": false,
|
| 54 |
+
"parent_class": "nn.LeakyReLU",
|
| 55 |
+
"quantizer_name": "*"
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"enable": false,
|
| 59 |
+
"quantizer_name": "*lm_head*"
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"enable": false,
|
| 63 |
+
"quantizer_name": "*proj_out.*"
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"enable": false,
|
| 67 |
+
"quantizer_name": "*block_sparse_moe.gate*"
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"enable": false,
|
| 71 |
+
"quantizer_name": "*router*"
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"enable": false,
|
| 75 |
+
"quantizer_name": "*mlp.gate.*"
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"enable": false,
|
| 79 |
+
"quantizer_name": "*mlp.shared_expert_gate.*"
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"enable": false,
|
| 83 |
+
"quantizer_name": "*linear_attn.conv1d*"
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"enable": false,
|
| 87 |
+
"quantizer_name": "*mixer.conv1d*"
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"enable": false,
|
| 91 |
+
"quantizer_name": "*output_layer*"
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"enable": false,
|
| 95 |
+
"quantizer_name": "output.*"
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"enable": false,
|
| 99 |
+
"quantizer_name": "*proj_in*"
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"enable": false,
|
| 103 |
+
"quantizer_name": "*proj_out*"
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"enable": false,
|
| 107 |
+
"quantizer_name": "*time_embedder*"
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"enable": false,
|
| 111 |
+
"quantizer_name": "*audio_proj_in*"
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"enable": false,
|
| 115 |
+
"quantizer_name": "*audio_proj_out*"
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"enable": false,
|
| 119 |
+
"quantizer_name": "*action_proj_in*"
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"enable": false,
|
| 123 |
+
"quantizer_name": "*action_proj_out*"
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"enable": false,
|
| 127 |
+
"quantizer_name": "*lm_head*"
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"enable": false,
|
| 131 |
+
"quantizer_name": "*visual*"
|
| 132 |
+
}
|
| 133 |
+
]
|
| 134 |
+
},
|
| 135 |
+
"modules_to_not_convert": [
|
| 136 |
+
"proj_in",
|
| 137 |
+
"proj_out",
|
| 138 |
+
"time_embedder",
|
| 139 |
+
"audio_proj_in",
|
| 140 |
+
"audio_proj_out",
|
| 141 |
+
"action_proj_in",
|
| 142 |
+
"action_proj_out",
|
| 143 |
+
"lm_head",
|
| 144 |
+
"visual"
|
| 145 |
+
],
|
| 146 |
+
"quant_method": "modelopt",
|
| 147 |
+
"quant_type": "FP8_FP8",
|
| 148 |
+
"scale_block_quantize": null,
|
| 149 |
+
"scale_channel_quantize": null,
|
| 150 |
+
"weight_only": false,
|
| 151 |
+
"config_groups": {
|
| 152 |
+
"group_0": {
|
| 153 |
+
"input_activations": {
|
| 154 |
+
"dynamic": false,
|
| 155 |
+
"num_bits": 8,
|
| 156 |
+
"type": "float"
|
| 157 |
+
},
|
| 158 |
+
"weights": {
|
| 159 |
+
"dynamic": false,
|
| 160 |
+
"num_bits": 8,
|
| 161 |
+
"type": "float"
|
| 162 |
+
},
|
| 163 |
+
"targets": [
|
| 164 |
+
"Linear"
|
| 165 |
+
]
|
| 166 |
+
}
|
| 167 |
+
},
|
| 168 |
+
"quant_algo": "FP8",
|
| 169 |
+
"ignore": [
|
| 170 |
+
"proj_in",
|
| 171 |
+
"proj_out",
|
| 172 |
+
"time_embedder*",
|
| 173 |
+
"audio_proj_in",
|
| 174 |
+
"audio_proj_out",
|
| 175 |
+
"action_proj_in",
|
| 176 |
+
"action_proj_out",
|
| 177 |
+
"lm_head",
|
| 178 |
+
"model.visual*",
|
| 179 |
+
"visual*"
|
| 180 |
+
],
|
| 181 |
+
"producer": {
|
| 182 |
+
"name": "modelopt",
|
| 183 |
+
"version": "0.44.0"
|
| 184 |
+
}
|
| 185 |
+
}
|
model.safetensors.index.json
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
transformer/config.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"_class_name": "Cosmos3OmniTransformer",
|
| 3 |
-
"_diffusers_version": "0.
|
| 4 |
"action_dim": 64,
|
| 5 |
"action_gen": true,
|
| 6 |
"attention_bias": false,
|
|
@@ -29,7 +29,218 @@
|
|
| 29 |
"qk_norm": false,
|
| 30 |
"qk_norm_for_diffusion": true,
|
| 31 |
"qk_norm_for_text": true,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
"rms_norm_eps": 1e-06,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
"rope_scaling": {
|
| 34 |
"mrope_interleaved": true,
|
| 35 |
"mrope_section": [
|
|
|
|
| 1 |
{
|
| 2 |
"_class_name": "Cosmos3OmniTransformer",
|
| 3 |
+
"_diffusers_version": "0.39.0",
|
| 4 |
"action_dim": 64,
|
| 5 |
"action_gen": true,
|
| 6 |
"attention_bias": false,
|
|
|
|
| 29 |
"qk_norm": false,
|
| 30 |
"qk_norm_for_diffusion": true,
|
| 31 |
"qk_norm_for_text": true,
|
| 32 |
+
"quantization_config": {
|
| 33 |
+
"block_quantize": null,
|
| 34 |
+
"calib_cfg": {
|
| 35 |
+
"method": "max"
|
| 36 |
+
},
|
| 37 |
+
"channel_quantize": null,
|
| 38 |
+
"disable_conv_quantization": false,
|
| 39 |
+
"forward_loop": null,
|
| 40 |
+
"modelopt_config": {
|
| 41 |
+
"algorithm": "max",
|
| 42 |
+
"quant_cfg": [
|
| 43 |
+
{
|
| 44 |
+
"enable": false,
|
| 45 |
+
"quantizer_name": "*"
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"cfg": {
|
| 49 |
+
"axis": null,
|
| 50 |
+
"fake_quant": false,
|
| 51 |
+
"num_bits": [
|
| 52 |
+
4,
|
| 53 |
+
3
|
| 54 |
+
]
|
| 55 |
+
},
|
| 56 |
+
"quantizer_name": "*weight_quantizer"
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"cfg": {
|
| 60 |
+
"axis": null,
|
| 61 |
+
"num_bits": [
|
| 62 |
+
4,
|
| 63 |
+
3
|
| 64 |
+
]
|
| 65 |
+
},
|
| 66 |
+
"quantizer_name": "*input_quantizer"
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"enable": false,
|
| 70 |
+
"parent_class": "nn.BatchNorm1d",
|
| 71 |
+
"quantizer_name": "*"
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"enable": false,
|
| 75 |
+
"parent_class": "nn.BatchNorm2d",
|
| 76 |
+
"quantizer_name": "*"
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"enable": false,
|
| 80 |
+
"parent_class": "nn.BatchNorm3d",
|
| 81 |
+
"quantizer_name": "*"
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"enable": false,
|
| 85 |
+
"parent_class": "nn.LeakyReLU",
|
| 86 |
+
"quantizer_name": "*"
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"enable": false,
|
| 90 |
+
"quantizer_name": "*lm_head*"
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"enable": false,
|
| 94 |
+
"quantizer_name": "*proj_out.*"
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"enable": false,
|
| 98 |
+
"quantizer_name": "*block_sparse_moe.gate*"
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"enable": false,
|
| 102 |
+
"quantizer_name": "*router*"
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"enable": false,
|
| 106 |
+
"quantizer_name": "*mlp.gate.*"
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"enable": false,
|
| 110 |
+
"quantizer_name": "*mlp.shared_expert_gate.*"
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"enable": false,
|
| 114 |
+
"quantizer_name": "*linear_attn.conv1d*"
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"enable": false,
|
| 118 |
+
"quantizer_name": "*mixer.conv1d*"
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"enable": false,
|
| 122 |
+
"quantizer_name": "*output_layer*"
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"enable": false,
|
| 126 |
+
"quantizer_name": "output.*"
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"enable": false,
|
| 130 |
+
"quantizer_name": "*proj_in*"
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"enable": false,
|
| 134 |
+
"quantizer_name": "*proj_out*"
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"enable": false,
|
| 138 |
+
"quantizer_name": "*time_embedder*"
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"enable": false,
|
| 142 |
+
"quantizer_name": "*audio_proj_in*"
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"enable": false,
|
| 146 |
+
"quantizer_name": "*audio_proj_out*"
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"enable": false,
|
| 150 |
+
"quantizer_name": "*action_proj_in*"
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"enable": false,
|
| 154 |
+
"quantizer_name": "*action_proj_out*"
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"enable": false,
|
| 158 |
+
"quantizer_name": "*lm_head*"
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"enable": false,
|
| 162 |
+
"quantizer_name": "*visual*"
|
| 163 |
+
}
|
| 164 |
+
]
|
| 165 |
+
},
|
| 166 |
+
"modules_to_not_convert": [
|
| 167 |
+
"proj_in",
|
| 168 |
+
"proj_out",
|
| 169 |
+
"time_embedder",
|
| 170 |
+
"audio_proj_in",
|
| 171 |
+
"audio_proj_out",
|
| 172 |
+
"action_proj_in",
|
| 173 |
+
"action_proj_out",
|
| 174 |
+
"lm_head",
|
| 175 |
+
"visual"
|
| 176 |
+
],
|
| 177 |
+
"quant_method": "modelopt",
|
| 178 |
+
"quant_type": "FP8_FP8",
|
| 179 |
+
"runtime": {
|
| 180 |
+
"diffusion_step_policy": {
|
| 181 |
+
"schema_version": 1,
|
| 182 |
+
"type": "first_last_n",
|
| 183 |
+
"index_space": "denoising_loop_iteration",
|
| 184 |
+
"scope": [
|
| 185 |
+
"transformer"
|
| 186 |
+
],
|
| 187 |
+
"default_mode": "native",
|
| 188 |
+
"first_steps": {
|
| 189 |
+
"count": 3,
|
| 190 |
+
"mode": "a16"
|
| 191 |
+
},
|
| 192 |
+
"last_steps": {
|
| 193 |
+
"count": 3,
|
| 194 |
+
"mode": "a16"
|
| 195 |
+
},
|
| 196 |
+
"overlap": "a16",
|
| 197 |
+
"reasoner": "a16"
|
| 198 |
+
}
|
| 199 |
+
},
|
| 200 |
+
"scale_block_quantize": null,
|
| 201 |
+
"scale_channel_quantize": null,
|
| 202 |
+
"weight_only": false,
|
| 203 |
+
"config_groups": {
|
| 204 |
+
"group_0": {
|
| 205 |
+
"input_activations": {
|
| 206 |
+
"dynamic": false,
|
| 207 |
+
"num_bits": 8,
|
| 208 |
+
"type": "float"
|
| 209 |
+
},
|
| 210 |
+
"weights": {
|
| 211 |
+
"dynamic": false,
|
| 212 |
+
"num_bits": 8,
|
| 213 |
+
"type": "float"
|
| 214 |
+
},
|
| 215 |
+
"targets": [
|
| 216 |
+
"Linear"
|
| 217 |
+
]
|
| 218 |
+
}
|
| 219 |
+
},
|
| 220 |
+
"quant_algo": "FP8",
|
| 221 |
+
"ignore": [
|
| 222 |
+
"proj_in",
|
| 223 |
+
"proj_out",
|
| 224 |
+
"time_embedder*",
|
| 225 |
+
"audio_proj_in",
|
| 226 |
+
"audio_proj_out",
|
| 227 |
+
"action_proj_in",
|
| 228 |
+
"action_proj_out",
|
| 229 |
+
"lm_head",
|
| 230 |
+
"model.visual*",
|
| 231 |
+
"visual*"
|
| 232 |
+
],
|
| 233 |
+
"producer": {
|
| 234 |
+
"name": "modelopt",
|
| 235 |
+
"version": "0.44.0"
|
| 236 |
+
}
|
| 237 |
+
},
|
| 238 |
"rms_norm_eps": 1e-06,
|
| 239 |
+
"rope_axes_dim": [
|
| 240 |
+
24,
|
| 241 |
+
20,
|
| 242 |
+
20
|
| 243 |
+
],
|
| 244 |
"rope_scaling": {
|
| 245 |
"mrope_interleaved": true,
|
| 246 |
"mrope_section": [
|
transformer/{diffusion_pytorch_model-00001-of-00027.safetensors → diffusion_pytorch_model-00001-of-00014.safetensors}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d585219f73c166577543b4bfa0bc069372cb704a271abec63cc1dc8b130220a3
|
| 3 |
+
size 4874978012
|
transformer/{diffusion_pytorch_model-00002-of-00027.safetensors → diffusion_pytorch_model-00002-of-00014.safetensors}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1d4b8bf159fd9acc71236de8e7b9f5d401a6e9edfdc691de301b09297c97ee3e
|
| 3 |
+
size 4975501784
|
transformer/{diffusion_pytorch_model-00003-of-00027.safetensors → diffusion_pytorch_model-00003-of-00014.safetensors}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:333e65aa5f832665b2403d600d355638c109a42c66975d3842be2d9ad6e63a20
|
| 3 |
+
size 4965490244
|
transformer/{diffusion_pytorch_model-00004-of-00027.safetensors → diffusion_pytorch_model-00004-of-00014.safetensors}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b13e0bf2995cfce2d27fa2baaeb6531069a3a80ace655bfd38eba6030504dee8
|
| 3 |
+
size 4959772504
|
transformer/diffusion_pytorch_model-00005-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e1d0bb56018e4ba34494989d06271765d3c626e28221c9e86a7780cf6ecac13
|
| 3 |
+
size 4996946796
|
transformer/diffusion_pytorch_model-00005-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:f4034914f825adcbda188584466a6c54ad8782bef1a88bafae117cf2d8aba2ab
|
| 3 |
-
size 4949368096
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00006-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cd0ca0c328ac1af236b8b8d63ff9b9681f08a4ec8d402008449c25b83c1465f8
|
| 3 |
+
size 4934076892
|
transformer/diffusion_pytorch_model-00006-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:dafd8015fcc8d7ae92c2246f5e742b1d7b5cfc098a6b64df6688312ec81555ab
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00007-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:15d1de9d96c1367bcbdaee5cf641e55f9c72a2f4010a3c3c22d241be7d01da31
|
| 3 |
+
size 4959772528
|
transformer/diffusion_pytorch_model-00007-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:4dc5a5e25a96a549b02ccc7f0d08a088c3eb93fa98651f83481d6822e9d2a534
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00008-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e325e5a5225b438b29f240ba4bad76cae7cb0ff7f9b911db522732f36500bb6e
|
| 3 |
+
size 4970258840
|
transformer/diffusion_pytorch_model-00008-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:25047e0b860d6ce87bb6e870578b3378d7697ae5e1c8a22068e0cd1bbc20ce9c
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00009-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e2a132915e7fc8e0a1954a4d4f205306aabed7622f702fcce855e0848388e152
|
| 3 |
+
size 4960249044
|
transformer/diffusion_pytorch_model-00009-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:8739245a796a498a6fab185d51632df3f4b15f37608573e270dddeb4ed4d9405
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00010-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c4f62cb7fd73bf0e8e01e5a289d449cd7706f3f9964d0e1d68edd155400b0c46
|
| 3 |
+
size 4975501856
|
transformer/diffusion_pytorch_model-00010-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:8e965d40a5c7c07d98c2554d48702a71d597c1ec31085cd10eee1d03f690310c
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00011-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63a61ac98d36c8422d771d066483562120d3cd1f64a096e1fdeb6e096dffca3a
|
| 3 |
+
size 4907817796
|
transformer/diffusion_pytorch_model-00011-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:bfb8aad1c0ac92d312e74c337e8506c9f274b72a914239ea6141194fd3d1438d
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00012-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e3684a8b8edb6fc496f839a65e280aed2aa1ad8b37dfb828d26e106ac15d2626
|
| 3 |
+
size 4975501856
|
transformer/diffusion_pytorch_model-00012-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:406e8c99fffef5e16bf4f2b3a512ccc347731227ad79a194d4ba875b95c8c6b9
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00013-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83a3ec1f9e4968e40df4689050fa7cfc13942addb03f5fc14ffd92944a6cb9f3
|
| 3 |
+
size 4877821680
|
transformer/diffusion_pytorch_model-00013-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:5acf40b6d7c3c97ceea8d53610588dfe3ecaac887321b3e8f38d8ea42a0d6375
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00014-of-00014.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3973097ad318754d065627f3dc36323563d6c9912fb7ca0de162eb8b4c31221
|
| 3 |
+
size 1295173056
|
transformer/diffusion_pytorch_model-00014-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:53c6562ab14a6158d1ac138760e85c3d1c3f4384e795cff6602731060259a5c5
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00015-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:b2208de377650a5afeec7e5d223ed5499459dd94b52aecb42df624f3be546a9f
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00016-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:987b6f80205f9b453996a1bbd8fc3fdca8ba6f59bd07675026cd37b7864929e8
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00017-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:d42b165e6c1448f415041b8d6a22a18b5a042e2714c90ba6ecafa1b460364234
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00018-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:8fe9c759b0b1126cc00671156959723bb427fd57288a293630ddb08a4e21c7af
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00019-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:89e3b04c0c39696a5b021d3a02a5816a5a82152f2b4fb53090734a79524a5a94
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00020-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:dd3f05dd899f97d682b850b9039db0ddd01fd63447581297c5f4299fe9ae7799
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00021-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:aafea01a1401f8335d20211ee3a7edc0453ce8943c6aeb36cc424f3b678ebdd8
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00022-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:42e4f9fbec544450716558e15a1437a359cd9f54e4d50106c2c061c1055564dd
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00023-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:41bf7a3c335767b869054b17374e094b062a6600c9912d8e03af886edb56a80b
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00024-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:9827a447db76396743e464c4088dca865a96819657c79dd367d8da53cf61e6a9
|
| 3 |
-
size 4802611024
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00025-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:a794c4f14a4442d49bdff78e33f10d611cc228c3e4b9383f4384972a7a71d430
|
| 3 |
-
size 4949368104
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00026-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:dc50ce14f37ddcd5fc5ea411d4ce475b11774bb8b013199a08341c0acc1e2273
|
| 3 |
-
size 4425141960
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model-00027-of-00027.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:11351800a5a691b6ddc17fc5d6376a459a853771814ed3b2533736fe9e9c6934
|
| 3 |
-
size 1658435704
|
|
|
|
|
|
|
|
|
|
|
|
transformer/diffusion_pytorch_model.safetensors.index.json
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
transformer/modelopt_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c4fa6a79f6312e6df0af0fee082b48004082d8533d54b972772faf19c6b8eebb
|
| 3 |
+
size 798423
|