Text-to-Image
Diffusers
Safetensors
MLX
cosmos3
quantization
int8
w8a16
sdnq
apple-silicon
cuda
8-bit precision
Instructions to use JuliaML/Cosmos3-Super-Text2Image-4Step-INT8-G64-BF16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use JuliaML/Cosmos3-Super-Text2Image-4Step-INT8-G64-BF16 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("JuliaML/Cosmos3-Super-Text2Image-4Step-INT8-G64-BF16", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - MLX
How to use JuliaML/Cosmos3-Super-Text2Image-4Step-INT8-G64-BF16 with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir Cosmos3-Super-Text2Image-4Step-INT8-G64-BF16 JuliaML/Cosmos3-Super-Text2Image-4Step-INT8-G64-BF16
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Draw Things
- DiffusionBee
- Atomic Chat
File size: 5,700 Bytes
23e8085 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 | {
"format": "cosmos3-sdnq-int8-verification-v1",
"source": "models/Cosmos3-Super-Text2Image-4Step/transformer",
"checkpoint": "checkpoints/Cosmos3-Super-Text2Image-4Step-SDNQ-INT8-G64-BF16IO/transformer",
"selection_report": "benchmarks/results/mac-grid-fix-priority/w8-common-selection-verification.json",
"weights_dtype": "int8",
"group_size": 64,
"selected_tensors": [
"layers.0.mlp.down_proj.weight",
"layers.10.self_attn.to_out.weight",
"layers.13.self_attn.add_v_proj.weight",
"layers.16.mlp_moe_gen.up_proj.weight",
"layers.19.mlp.up_proj.weight",
"layers.20.self_attn.to_v.weight",
"layers.23.self_attn.to_add_out.weight",
"layers.26.self_attn.add_k_proj.weight",
"layers.29.mlp_moe_gen.down_proj.weight",
"layers.31.mlp.down_proj.weight",
"layers.33.self_attn.to_out.weight",
"layers.36.self_attn.add_v_proj.weight",
"layers.39.mlp_moe_gen.up_proj.weight",
"layers.41.mlp.up_proj.weight",
"layers.43.self_attn.to_v.weight",
"layers.46.self_attn.to_k.weight",
"layers.49.self_attn.add_q_proj.weight",
"layers.51.mlp_moe_gen.gate_proj.weight",
"layers.54.mlp.down_proj.weight",
"layers.56.self_attn.to_out.weight",
"layers.59.self_attn.add_v_proj.weight",
"layers.61.mlp_moe_gen.up_proj.weight",
"layers.7.mlp.up_proj.weight",
"layers.9.self_attn.to_v.weight",
"proj_in.weight",
"proj_out.weight"
],
"packing_exact": true,
"global_relative_mse": 3.8788798613407266e-05,
"core_global_relative_mse": 3.8788798613407266e-05,
"elapsed_seconds": 2.4776670410001316,
"tensors": [
{
"name": "layers.0.mlp.down_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 4.137296453991857e-05
},
{
"name": "layers.10.self_attn.to_out.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.550073523138151e-05
},
{
"name": "layers.13.self_attn.add_v_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 6.126165909311447e-05
},
{
"name": "layers.16.mlp_moe_gen.up_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 5.150241390340871e-05
},
{
"name": "layers.19.mlp.up_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.6130995357940006e-05
},
{
"name": "layers.20.self_attn.to_v.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.731611253132676e-05
},
{
"name": "layers.23.self_attn.to_add_out.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.4582748715957426e-05
},
{
"name": "layers.26.self_attn.add_k_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 4.003427992115351e-05
},
{
"name": "layers.29.mlp_moe_gen.down_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.823720056398718e-05
},
{
"name": "layers.31.mlp.down_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.7567574480100976e-05
},
{
"name": "layers.33.self_attn.to_out.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.549189601147462e-05
},
{
"name": "layers.36.self_attn.add_v_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.770933684528822e-05
},
{
"name": "layers.39.mlp_moe_gen.up_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.84747254049213e-05
},
{
"name": "layers.41.mlp.up_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.7791982573130374e-05
},
{
"name": "layers.43.self_attn.to_v.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.935090356148701e-05
},
{
"name": "layers.46.self_attn.to_k.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 4.075648970728934e-05
},
{
"name": "layers.49.self_attn.add_q_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 4.0976243349428724e-05
},
{
"name": "layers.51.mlp_moe_gen.gate_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 4.32475790259716e-05
},
{
"name": "layers.54.mlp.down_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.6751653390400224e-05
},
{
"name": "layers.56.self_attn.to_out.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.558971994482112e-05
},
{
"name": "layers.59.self_attn.add_v_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.7726695875040836e-05
},
{
"name": "layers.61.mlp_moe_gen.up_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.915807374858549e-05
},
{
"name": "layers.7.mlp.up_proj.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.673001616531715e-05
},
{
"name": "layers.9.self_attn.to_v.weight",
"quantized": true,
"packing_exact": true,
"relative_mse": 3.710549181245947e-05
},
{
"name": "proj_in.weight",
"quantized": false,
"exact": true
},
{
"name": "proj_out.weight",
"quantized": false,
"exact": true
}
]
}
|