Hugging Face's logo

JuliaML
/
Cosmos3-Super-Image2Video-4Step-INT8-G64-BF16

Image-to-Video
Diffusers
Safetensors
MLX
cosmos3
video-generation
quantization
int8
w8a16
sdnq
apple-silicon
cuda
8-bit precision
Model card Files
xet
 Community
Cosmos3-Super-Image2Video-4Step-INT8-G64-BF16
File size: 2,604 Bytes
89f2fc0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
{
  "format": "cosmos3-sdnq-verification-v1",
  "source": "models/Cosmos3-Super-Image2Video-4Step/transformer",
  "checkpoint": "checkpoints/Cosmos3-Super-Image2Video-4Step-SDNQ-INT8-G64-BF16IO/transformer",
  "selection_report": "benchmarks/i2v/verification-selection.json",
  "weights_dtype": "int8",
  "group_size": 64,
  "selected_tensors": [
    "embed_tokens.weight",
    "layers.0.self_attn.to_q.weight",
    "layers.0.mlp.up_proj.weight",
    "layers.21.self_attn.to_k.weight",
    "layers.21.mlp_moe_gen.down_proj.weight",
    "layers.42.self_attn.add_v_proj.weight",
    "layers.42.mlp.gate_proj.weight",
    "layers.63.self_attn.to_out.weight",
    "layers.63.mlp.down_proj.weight",
    "norm.weight",
    "proj_in.weight",
    "proj_out.weight"
  ],
  "packing_exact": true,
  "global_relative_mse": 4.091530100791147e-05,
  "core_global_relative_mse": 4.091530100791147e-05,
  "elapsed_seconds": 2.059502917000003,
  "tensors": [
    {
      "name": "embed_tokens.weight",
      "quantized": false,
      "exact": true
    },
    {
      "name": "layers.0.self_attn.to_q.weight",
      "quantized": true,
      "packing_exact": true,
      "relative_mse": 3.631787930218696e-05
    },
    {
      "name": "layers.0.mlp.up_proj.weight",
      "quantized": true,
      "packing_exact": true,
      "relative_mse": 3.787928616525752e-05
    },
    {
      "name": "layers.21.self_attn.to_k.weight",
      "quantized": true,
      "packing_exact": true,
      "relative_mse": 4.015870025460336e-05
    },
    {
      "name": "layers.21.mlp_moe_gen.down_proj.weight",
      "quantized": true,
      "packing_exact": true,
      "relative_mse": 4.617328465040616e-05
    },
    {
      "name": "layers.42.self_attn.add_v_proj.weight",
      "quantized": true,
      "packing_exact": true,
      "relative_mse": 4.422963696925805e-05
    },
    {
      "name": "layers.42.mlp.gate_proj.weight",
      "quantized": true,
      "packing_exact": true,
      "relative_mse": 3.8449613650290844e-05
    },
    {
      "name": "layers.63.self_attn.to_out.weight",
      "quantized": true,
      "packing_exact": true,
      "relative_mse": 3.570206942213908e-05
    },
    {
      "name": "layers.63.mlp.down_proj.weight",
      "quantized": true,
      "packing_exact": true,
      "relative_mse": 4.066047565003439e-05
    },
    {
      "name": "norm.weight",
      "quantized": false,
      "exact": true
    },
    {
      "name": "proj_in.weight",
      "quantized": false,
      "exact": true
    },
    {
      "name": "proj_out.weight",
      "quantized": false,
      "exact": true
    }
  ]
}