Xenova HF staff commited on
Commit
c7e79b1
1 Parent(s): bda595e

Upload folder using huggingface_hub

Browse files
config.json CHANGED
@@ -3,6 +3,7 @@
3
  "architectures": [
4
  "T5ForConditionalGeneration"
5
  ],
 
6
  "d_ff": 2048,
7
  "d_kv": 64,
8
  "d_model": 512,
@@ -53,7 +54,7 @@
53
  "prefix": "translate English to Romanian: "
54
  }
55
  },
56
- "transformers_version": "4.29.2",
57
  "use_cache": true,
58
  "vocab_size": 32128
59
  }
 
3
  "architectures": [
4
  "T5ForConditionalGeneration"
5
  ],
6
+ "classifier_dropout": 0.0,
7
  "d_ff": 2048,
8
  "d_kv": 64,
9
  "d_model": 512,
 
54
  "prefix": "translate English to Romanian: "
55
  }
56
  },
57
+ "transformers_version": "4.34.0.dev0",
58
  "use_cache": true,
59
  "vocab_size": 32128
60
  }
generation_config.json CHANGED
@@ -3,5 +3,5 @@
3
  "decoder_start_token_id": 0,
4
  "eos_token_id": 1,
5
  "pad_token_id": 0,
6
- "transformers_version": "4.29.2"
7
  }
 
3
  "decoder_start_token_id": 0,
4
  "eos_token_id": 1,
5
  "pad_token_id": 0,
6
+ "transformers_version": "4.34.0.dev0"
7
  }
onnx/decoder_model.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3a0cb4929ba7e898e908f9f0f9124bb0ab81555a49e9c3de43fc411168638f71
3
- size 232468834
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e46abb87e26d82ced401c78105b3d7b254f890800f5ef224688811de171255c
3
+ size 166670419
onnx/decoder_model_merged.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7852d50d3d56481cbb19275dadacd11b9847e233eea7248816ee1c825dbc198d
3
- size 232631325
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:09ce39b5dad1b89b81cf068f3046f0e590f36510319c6eecec81163f0f4b5bf8
3
+ size 166832597
onnx/decoder_model_merged_quantized.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:81232f7c9e9f7630b9ddc7189b1d1a07de3fe2f98c7c678a143c94089ad5a2d0
3
- size 59043178
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:468bb04cc7cbdf77881560c9876014c87b7ccc00d53521924d6568aa369d162e
3
+ size 42431406
onnx/decoder_model_quantized.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9fac0e0a3fbe5d13e6cf884066d597ee9c38791f63baa4d1798d54a68c886e27
3
- size 58765648
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a27ad523566782fec8b429c1049e1707a19905563b685c89bf4b6baac27a4d1
3
+ size 42154654
onnx/decoder_with_past_model.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:71c48b8cb7e1a6f7e5bb415a7781449fc318d1ed443580ff29a1c3ec9628b6fe
3
- size 219874970
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fffa272f238f440e69fe2bf0408b628974a4e177bae114f33ebf4e8b66641aa5
3
+ size 154076559
onnx/decoder_with_past_model_quantized.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:49d62404830490bd0d8e0b422839f0ed152286e0f2682c55711560aa90c5c8ec
3
- size 55563101
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d85143968e6d06590e3c1a907996d3d6f0d610126988f365529151460425b79
3
+ size 38952121
quantize_config.json ADDED
@@ -0,0 +1,139 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_channel": true,
3
+ "reduce_range": true,
4
+ "per_model_config": {
5
+ "decoder_model": {
6
+ "op_types": [
7
+ "Pow",
8
+ "Cast",
9
+ "Sub",
10
+ "Concat",
11
+ "ReduceMean",
12
+ "Unsqueeze",
13
+ "Shape",
14
+ "Div",
15
+ "ConstantOfShape",
16
+ "Log",
17
+ "Constant",
18
+ "Sqrt",
19
+ "Min",
20
+ "LessOrEqual",
21
+ "Neg",
22
+ "Transpose",
23
+ "Reshape",
24
+ "Range",
25
+ "Tile",
26
+ "Mul",
27
+ "Where",
28
+ "Relu",
29
+ "Softmax",
30
+ "MatMul",
31
+ "Gather",
32
+ "Less",
33
+ "Expand",
34
+ "Add"
35
+ ],
36
+ "weight_type": "QInt8"
37
+ },
38
+ "encoder_model": {
39
+ "op_types": [
40
+ "Pow",
41
+ "Cast",
42
+ "Sub",
43
+ "Concat",
44
+ "ReduceMean",
45
+ "Unsqueeze",
46
+ "Shape",
47
+ "Div",
48
+ "Greater",
49
+ "ConstantOfShape",
50
+ "Log",
51
+ "Constant",
52
+ "Sqrt",
53
+ "Min",
54
+ "Transpose",
55
+ "Reshape",
56
+ "Mul",
57
+ "Range",
58
+ "Abs",
59
+ "Where",
60
+ "Relu",
61
+ "Softmax",
62
+ "MatMul",
63
+ "Gather",
64
+ "Less",
65
+ "Add"
66
+ ],
67
+ "weight_type": "QInt8"
68
+ },
69
+ "decoder_model_merged": {
70
+ "op_types": [
71
+ "Pow",
72
+ "Cast",
73
+ "Sub",
74
+ "Concat",
75
+ "ReduceMean",
76
+ "Unsqueeze",
77
+ "Shape",
78
+ "Div",
79
+ "ConstantOfShape",
80
+ "Log",
81
+ "Slice",
82
+ "Constant",
83
+ "Sqrt",
84
+ "Min",
85
+ "If",
86
+ "LessOrEqual",
87
+ "Neg",
88
+ "Transpose",
89
+ "Reshape",
90
+ "Range",
91
+ "Tile",
92
+ "Mul",
93
+ "Where",
94
+ "Relu",
95
+ "Softmax",
96
+ "MatMul",
97
+ "Gather",
98
+ "Less",
99
+ "Expand",
100
+ "Add"
101
+ ],
102
+ "weight_type": "QInt8"
103
+ },
104
+ "decoder_with_past_model": {
105
+ "op_types": [
106
+ "Pow",
107
+ "Cast",
108
+ "Sub",
109
+ "Concat",
110
+ "ReduceMean",
111
+ "Unsqueeze",
112
+ "Shape",
113
+ "Div",
114
+ "ConstantOfShape",
115
+ "Log",
116
+ "Slice",
117
+ "Constant",
118
+ "Sqrt",
119
+ "Min",
120
+ "LessOrEqual",
121
+ "Neg",
122
+ "Transpose",
123
+ "Reshape",
124
+ "Range",
125
+ "Tile",
126
+ "Mul",
127
+ "Where",
128
+ "Relu",
129
+ "Softmax",
130
+ "MatMul",
131
+ "Gather",
132
+ "Less",
133
+ "Expand",
134
+ "Add"
135
+ ],
136
+ "weight_type": "QInt8"
137
+ }
138
+ }
139
+ }
tokenizer_config.json CHANGED
@@ -104,8 +104,12 @@
104
  "clean_up_tokenization_spaces": true,
105
  "eos_token": "</s>",
106
  "extra_ids": 100,
 
107
  "model_max_length": 512,
108
  "pad_token": "<pad>",
 
109
  "tokenizer_class": "T5Tokenizer",
 
 
110
  "unk_token": "<unk>"
111
  }
 
104
  "clean_up_tokenization_spaces": true,
105
  "eos_token": "</s>",
106
  "extra_ids": 100,
107
+ "max_length": 512,
108
  "model_max_length": 512,
109
  "pad_token": "<pad>",
110
+ "stride": 0,
111
  "tokenizer_class": "T5Tokenizer",
112
+ "truncation_side": "right",
113
+ "truncation_strategy": "longest_first",
114
  "unk_token": "<unk>"
115
  }