Xenova HF staff commited on
Commit
1c7ebbb
1 Parent(s): f56109e

Upload folder using huggingface_hub

Browse files
onnx/decoder_model_merged.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1b0f375b96dd3a76b25f4d5377eda6376faae560682bed258510f0bbd8289470
3
  size 791453859
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55fbbfbdd774d0f4cc2a92f59f11e58584539d3a92dfb6536f65d4d3c1c4f57a
3
  size 791453859
onnx/decoder_model_merged_quantized.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:be7061281a908c61b62bd0bc9dea7a707a72d55a374deab3ccfaec173acc2aa2
3
- size 299954775
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4699d4bb1950bb65f77e2f6ce1b26a51f26fdb5813e22581efcf0c40fd8b95b4
3
+ size 300505575
onnx/decoder_model_quantized.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6889a8fdc9daf18b7aa3c6124684eaffd0fb2fba6ee0103064ad719d79d7bf0a
3
- size 232306103
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a960ff52fcdff888c99beb87c7a497165a3e2201b3c24777ce295b37c546062
3
+ size 232856903
onnx/decoder_with_past_model_quantized.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:66061e6a38529a523db921f29bebea1086e4753e60daff773ecfe8fb48ad0b10
3
- size 232315060
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c964f2c88afb32affe47ab3d531c48617ebe071b393ad5eec09c1e410948b4bf
3
+ size 232865860
quantize_config.json CHANGED
@@ -1,98 +1,98 @@
1
  {
2
- "per_channel": false,
3
- "reduce_range": false,
4
  "per_model_config": {
5
  "decoder_model": {
6
  "op_types": [
 
 
 
 
 
7
  "Add",
8
- "Shape",
9
- "Where",
10
- "Sub",
11
- "Unsqueeze",
12
- "Equal",
13
- "Concat",
14
- "Reshape",
15
- "Softmax",
16
  "Constant",
17
- "Tanh",
18
- "Gather",
19
- "CumSum",
20
- "Squeeze",
21
- "Sqrt",
22
- "Cast",
23
- "Slice",
24
  "Split",
25
- "Transpose",
 
26
  "ConstantOfShape",
27
- "Mul",
28
- "Div",
 
 
 
 
29
  "MatMul",
30
- "And",
31
- "Pow",
32
- "ReduceMean"
 
 
33
  ],
34
  "weight_type": "QInt8"
35
  },
36
  "decoder_model_merged": {
37
  "op_types": [
 
 
 
 
 
38
  "Add",
39
- "Shape",
40
- "Where",
41
- "Sub",
42
- "Unsqueeze",
43
- "Equal",
44
- "Concat",
45
- "Reshape",
46
- "Softmax",
47
  "Constant",
48
- "Tanh",
49
- "Gather",
50
- "CumSum",
51
- "Squeeze",
52
- "Sqrt",
53
- "Cast",
54
- "Slice",
55
  "Split",
56
- "Transpose",
 
57
  "ConstantOfShape",
58
- "If",
59
- "Mul",
60
- "Div",
 
 
 
61
  "MatMul",
62
- "And",
63
- "Pow",
64
- "ReduceMean"
 
 
65
  ],
66
  "weight_type": "QInt8"
67
  },
68
  "decoder_with_past_model": {
69
  "op_types": [
 
 
 
 
 
70
  "Add",
71
- "Shape",
72
- "Where",
73
- "Sub",
74
- "Unsqueeze",
75
- "Equal",
76
- "Concat",
77
- "Reshape",
78
- "Softmax",
79
  "Constant",
80
- "Tanh",
81
- "Gather",
82
- "Squeeze",
83
- "CumSum",
84
- "Sqrt",
85
- "Slice",
86
- "Cast",
87
  "Split",
88
- "Transpose",
 
89
  "ConstantOfShape",
90
- "Mul",
91
- "Div",
 
 
 
 
92
  "MatMul",
93
- "And",
94
- "Pow",
95
- "ReduceMean"
 
 
96
  ],
97
  "weight_type": "QInt8"
98
  }
 
1
  {
2
+ "per_channel": true,
3
+ "reduce_range": true,
4
  "per_model_config": {
5
  "decoder_model": {
6
  "op_types": [
7
+ "Transpose",
8
+ "ReduceMean",
9
+ "Div",
10
+ "Gather",
11
+ "And",
12
  "Add",
 
 
 
 
 
 
 
 
13
  "Constant",
14
+ "Mul",
15
+ "Concat",
16
+ "Sub",
 
 
 
 
17
  "Split",
18
+ "Pow",
19
+ "Tanh",
20
  "ConstantOfShape",
21
+ "Softmax",
22
+ "Cast",
23
+ "Sqrt",
24
+ "Equal",
25
+ "Squeeze",
26
+ "Shape",
27
  "MatMul",
28
+ "Where",
29
+ "Reshape",
30
+ "Unsqueeze",
31
+ "Slice",
32
+ "CumSum"
33
  ],
34
  "weight_type": "QInt8"
35
  },
36
  "decoder_model_merged": {
37
  "op_types": [
38
+ "Transpose",
39
+ "ReduceMean",
40
+ "Div",
41
+ "Gather",
42
+ "And",
43
  "Add",
 
 
 
 
 
 
 
 
44
  "Constant",
45
+ "Mul",
46
+ "Concat",
47
+ "If",
48
+ "Sub",
 
 
 
49
  "Split",
50
+ "Pow",
51
+ "Tanh",
52
  "ConstantOfShape",
53
+ "Softmax",
54
+ "Cast",
55
+ "Sqrt",
56
+ "Equal",
57
+ "Squeeze",
58
+ "Shape",
59
  "MatMul",
60
+ "Where",
61
+ "Reshape",
62
+ "Unsqueeze",
63
+ "Slice",
64
+ "CumSum"
65
  ],
66
  "weight_type": "QInt8"
67
  },
68
  "decoder_with_past_model": {
69
  "op_types": [
70
+ "Transpose",
71
+ "ReduceMean",
72
+ "Div",
73
+ "Gather",
74
+ "And",
75
  "Add",
 
 
 
 
 
 
 
 
76
  "Constant",
77
+ "Mul",
78
+ "Concat",
79
+ "Sub",
 
 
 
 
80
  "Split",
81
+ "Pow",
82
+ "Tanh",
83
  "ConstantOfShape",
84
+ "Softmax",
85
+ "Cast",
86
+ "Sqrt",
87
+ "Equal",
88
+ "Squeeze",
89
+ "Shape",
90
  "MatMul",
91
+ "Where",
92
+ "Reshape",
93
+ "Unsqueeze",
94
+ "Slice",
95
+ "CumSum"
96
  ],
97
  "weight_type": "QInt8"
98
  }