openvino-ci commited on
Commit
847a051
1 Parent(s): a9127c9

Upload folder using huggingface_hub

Browse files
README.md CHANGED
@@ -1,39 +1,37 @@
1
  ---
2
  license: apache-2.0
3
- language:
4
- - en
5
  ---
6
-
7
  # codegen25-7b-multi-int8-ov
8
-
9
- * Model creator: [Salesforce](https://huggingface.co/Salesforce)
10
- * Original model: [CodeGen2.5-7B-multi](https://huggingface.co/Salesforce/codegen25-7b-multi_P)
11
 
12
  ## Description
13
-
14
- This is [CodeGen2.5-7B-multi](https://huggingface.co/Salesforce/codegen25-7b-multi_P) model converted to the [OpenVINO™ IR](https://docs.openvino.ai/2024/documentation/openvino-ir-format.html) (Intermediate Representation) format with weights compressed to FP16.
15
 
16
  ## Quantization Parameters
17
 
18
  Weight compression was performed using `nncf.compress_weights` with the following parameters:
19
 
20
- * mode: **INT8_ASYM**
 
21
 
22
  For more information on quantization, check the [OpenVINO model optimization guide](https://docs.openvino.ai/2024/openvino-workflow/model-optimization-guide/weight-compression.html).
23
 
 
24
  ## Compatibility
25
 
26
  The provided OpenVINO™ IR model is compatible with:
27
 
28
- * OpenVINO version 2024.2.0 and higher
29
- * Optimum Intel 1.17.0 and higher
30
 
31
- ## Running Model Inference with [Optimum Intel](https://huggingface.co/docs/optimum/intel/index)
32
 
33
  1. Install packages required for using [Optimum Intel](https://huggingface.co/docs/optimum/intel/index) integration with the OpenVINO backend:
34
 
35
  ```
36
- pip install optimum[openvino] tiktoken
37
  ```
38
 
39
  2. Run model inference:
@@ -43,55 +41,25 @@ from transformers import AutoTokenizer
43
  from optimum.intel.openvino import OVModelForCausalLM
44
 
45
  model_id = "OpenVINO/codegen25-7b-multi-int8-ov"
46
- tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
47
  model = OVModelForCausalLM.from_pretrained(model_id)
48
- text = "def hello_world():"
49
- input_ids = tokenizer(text, return_tensors="pt").input_ids
50
- generated_ids = model.generate(input_ids, max_length=128)
51
- print(tokenizer.decode(generated_ids[0], skip_special_tokens=True))
52
- ```
53
-
54
- For more examples and possible optimizations, refer to the [OpenVINO Large Language Model Inference Guide](https://docs.openvino.ai/2024/learn-openvino/llm_inference_guide.html).
55
-
56
- ## Running Model Inference with [OpenVINO GenAI](https://github.com/openvinotoolkit/openvino.genai)
57
-
58
- 1. Install packages required for using OpenVINO GenAI.
59
- ```
60
- pip install openvino-genai huggingface_hub
61
- ```
62
 
63
- 2. Download model from HuggingFace Hub
64
-
65
- ```
66
- import huggingface_hub as hf_hub
67
-
68
- model_id = "OpenVINO/codegen25-7b-multi-int8-ov"
69
- model_path = "codegen25-7b-multi-int8-ov"
70
-
71
- hf_hub.snapshot_download(model_id, local_dir=model_path)
72
 
 
 
 
73
  ```
74
 
75
- 3. Run model inference:
76
-
77
- ```
78
- import openvino_genai as ov_genai
79
-
80
- device = "CPU"
81
- pipe = ov_genai.LLMPipeline(model_path, device)
82
- print(pipe.generate("def hello_world():", max_length=200))
83
- ```
84
-
85
- More GenAI usage examples can be found in OpenVINO GenAI library [docs](https://github.com/openvinotoolkit/openvino.genai/blob/master/src/README.md) and [samples](https://github.com/openvinotoolkit/openvino.genai?tab=readme-ov-file#openvino-genai-samples)
86
-
87
 
88
  ## Limitations
89
 
90
- Check the original model card for [limitations](https://huggingface.co/Salesforce/codegen25-7b-instruct_P#intended-use-and-limitations).
91
 
92
  ## Legal information
93
 
94
- The original model is distributed under [Apache 2.0](https://choosealicense.com/licenses/apache-2.0/) license. More details can be found in [original model card](https://huggingface.co/Salesforce/codegen25-7b-multi_P).
95
 
96
  ## Disclaimer
97
 
 
1
  ---
2
  license: apache-2.0
3
+ license_link: https://choosealicense.com/licenses/apache-2.0/
 
4
  ---
 
5
  # codegen25-7b-multi-int8-ov
6
+ * Model creator: [Salesforce](https://huggingface.co/Salesforce)
7
+ * Original model: [codegen25-7b-multi](https://huggingface.co/Salesforce/codegen25-7b-multi)
 
8
 
9
  ## Description
10
+ This is [codegen25-7b-multi](https://huggingface.co/Salesforce/codegen25-7b-multi) model converted to the [OpenVINO™ IR](https://docs.openvino.ai/2024/documentation/openvino-ir-format.html) (Intermediate Representation) format with weights compressed to INT8 by [NNCF](https://github.com/openvinotoolkit/nncf).
 
11
 
12
  ## Quantization Parameters
13
 
14
  Weight compression was performed using `nncf.compress_weights` with the following parameters:
15
 
16
+ * mode: **int8_asym**
17
+ * ratio: **1**
18
 
19
  For more information on quantization, check the [OpenVINO model optimization guide](https://docs.openvino.ai/2024/openvino-workflow/model-optimization-guide/weight-compression.html).
20
 
21
+
22
  ## Compatibility
23
 
24
  The provided OpenVINO™ IR model is compatible with:
25
 
26
+ * OpenVINO version 2024.4.0 and higher
27
+ * Optimum Intel 1.20.0 and higher
28
 
29
+ ## Running Model Inference
30
 
31
  1. Install packages required for using [Optimum Intel](https://huggingface.co/docs/optimum/intel/index) integration with the OpenVINO backend:
32
 
33
  ```
34
+ pip install optimum[openvino]
35
  ```
36
 
37
  2. Run model inference:
 
41
  from optimum.intel.openvino import OVModelForCausalLM
42
 
43
  model_id = "OpenVINO/codegen25-7b-multi-int8-ov"
44
+ tokenizer = AutoTokenizer.from_pretrained(model_id)
45
  model = OVModelForCausalLM.from_pretrained(model_id)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
46
 
47
+ inputs = tokenizer("What is OpenVINO?", return_tensors="pt")
 
 
 
 
 
 
 
 
48
 
49
+ outputs = model.generate(**inputs, max_length=200)
50
+ text = tokenizer.batch_decode(outputs)[0]
51
+ print(text)
52
  ```
53
 
54
+ For more examples and possible optimizations, refer to the [OpenVINO Large Language Model Inference Guide](https://docs.openvino.ai/2024/learn-openvino/llm_inference_guide.html).
 
 
 
 
 
 
 
 
 
 
 
55
 
56
  ## Limitations
57
 
58
+ Check the original model card for [original model card](https://huggingface.co/Salesforce/codegen25-7b-multi) for limitations.
59
 
60
  ## Legal information
61
 
62
+ The original model is distributed under [apache-2.0](https://choosealicense.com/licenses/apache-2.0/) license. More details can be found in [original model card](https://huggingface.co/Salesforce/codegen25-7b-multi).
63
 
64
  ## Disclaimer
65
 
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "Salesforce/codegen25-7b-multi_P",
3
  "architectures": [
4
  "LlamaForCausalLM"
5
  ],
@@ -7,11 +7,13 @@
7
  "attention_dropout": 0.0,
8
  "bos_token_id": 50256,
9
  "eos_token_id": 50256,
 
10
  "hidden_act": "silu",
11
  "hidden_size": 4096,
12
  "initializer_range": 0.02,
13
  "intermediate_size": 11008,
14
  "max_position_embeddings": 2048,
 
15
  "model_type": "llama",
16
  "num_attention_heads": 32,
17
  "num_hidden_layers": 32,
@@ -22,7 +24,7 @@
22
  "rope_scaling": null,
23
  "rope_theta": 10000.0,
24
  "tie_word_embeddings": false,
25
- "transformers_version": "4.40.1",
26
  "use_cache": true,
27
  "vocab_size": 51200
28
- }
 
1
  {
2
+ "_name_or_path": "Salesforce/codegen25-7b-multi",
3
  "architectures": [
4
  "LlamaForCausalLM"
5
  ],
 
7
  "attention_dropout": 0.0,
8
  "bos_token_id": 50256,
9
  "eos_token_id": 50256,
10
+ "head_dim": 128,
11
  "hidden_act": "silu",
12
  "hidden_size": 4096,
13
  "initializer_range": 0.02,
14
  "intermediate_size": 11008,
15
  "max_position_embeddings": 2048,
16
+ "mlp_bias": false,
17
  "model_type": "llama",
18
  "num_attention_heads": 32,
19
  "num_hidden_layers": 32,
 
24
  "rope_scaling": null,
25
  "rope_theta": 10000.0,
26
  "tie_word_embeddings": false,
27
+ "transformers_version": "4.45.2",
28
  "use_cache": true,
29
  "vocab_size": 51200
30
+ }
generation_config.json CHANGED
@@ -2,5 +2,5 @@
2
  "_from_model_config": true,
3
  "bos_token_id": 50256,
4
  "eos_token_id": 50256,
5
- "transformers_version": "4.40.1"
6
  }
 
2
  "_from_model_config": true,
3
  "bos_token_id": 50256,
4
  "eos_token_id": 50256,
5
+ "transformers_version": "4.45.2"
6
  }
openvino_detokenizer.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0669ffcbd41b89ee947a59aa77dafa089d57a808d9421da5751cd147e064f329
3
- size 572328
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:17900eba810d745a2800e475e41bff8587f51217883ee7680fa3f3000bafdfaf
3
+ size 535202
openvino_detokenizer.xml CHANGED
@@ -1,16 +1,16 @@
1
  <?xml version="1.0"?>
2
  <net name="detokenizer" version="11">
3
  <layers>
4
- <layer id="0" name="Parameter_186" type="Parameter" version="opset1">
5
  <data shape="?,?" element_type="i64" />
6
  <output>
7
- <port id="0" precision="I64" names="Parameter_186">
8
  <dim>-1</dim>
9
  <dim>-1</dim>
10
  </port>
11
  </output>
12
  </layer>
13
- <layer id="1" name="Convert_202" type="Convert" version="opset1">
14
  <data destination_type="i32" />
15
  <input>
16
  <port id="0" precision="I64">
@@ -25,19 +25,19 @@
25
  </port>
26
  </output>
27
  </layer>
28
- <layer id="2" name="Constant_89" type="Const" version="opset1">
29
- <data element_type="u8" shape="572279" offset="0" size="572279" />
30
  <output>
31
  <port id="0" precision="U8">
32
- <dim>572279</dim>
33
  </port>
34
  </output>
35
  </layer>
36
- <layer id="3" name="StringTensorUnpack_90" type="StringTensorUnpack" version="extension">
37
  <data mode="begins_ends" />
38
  <input>
39
  <port id="0" precision="U8">
40
- <dim>572279</dim>
41
  </port>
42
  </input>
43
  <output>
@@ -52,7 +52,7 @@
52
  </port>
53
  </output>
54
  </layer>
55
- <layer id="4" name="VocabDecoder_187" type="VocabDecoder" version="extension">
56
  <data skip_tokens="50256" />
57
  <input>
58
  <port id="0" precision="I32">
@@ -87,7 +87,7 @@
87
  </port>
88
  </output>
89
  </layer>
90
- <layer id="5" name="CharsToBytes_188" type="CharsToBytes" version="extension">
91
  <input>
92
  <port id="0" precision="I32">
93
  <dim>-1</dim>
@@ -101,39 +101,33 @@
101
  <port id="3" precision="I32">
102
  <dim>-1</dim>
103
  </port>
104
- <port id="4" precision="U8">
105
- <dim>-1</dim>
106
- </port>
107
  </input>
108
  <output>
109
- <port id="5" precision="I32">
110
- <dim>-1</dim>
111
- </port>
112
- <port id="6" precision="I32">
113
  <dim>-1</dim>
114
  </port>
115
- <port id="7" precision="U8">
116
  <dim>-1</dim>
117
  </port>
118
  </output>
119
  </layer>
120
- <layer id="6" name="Constant_190" type="Const" version="opset1">
121
- <data element_type="u8" shape="47" offset="572279" size="47" />
122
  <output>
123
  <port id="0" precision="U8">
124
  <dim>47</dim>
125
  </port>
126
  </output>
127
  </layer>
128
- <layer id="7" name="Constant_192" type="Const" version="opset1">
129
- <data element_type="u8" shape="2" offset="572326" size="2" />
130
  <output>
131
  <port id="0" precision="U8">
132
  <dim>2</dim>
133
  </port>
134
  </output>
135
  </layer>
136
- <layer id="8" name="RegexNormalization_193" type="RegexNormalization" version="extension">
137
  <data global_replace="true" />
138
  <input>
139
  <port id="0" precision="I32">
@@ -164,7 +158,7 @@
164
  </port>
165
  </output>
166
  </layer>
167
- <layer id="9" name="StringTensorPack_194" type="StringTensorPack" version="extension">
168
  <data mode="begins_ends" />
169
  <input>
170
  <port id="0" precision="I32">
@@ -183,7 +177,7 @@
183
  </port>
184
  </output>
185
  </layer>
186
- <layer id="10" name="Result_195" type="Result" version="opset1">
187
  <input>
188
  <port id="0" precision="STRING">
189
  <dim>-1</dim>
@@ -198,14 +192,13 @@
198
  <edge from-layer="3" from-port="1" to-layer="4" to-port="1" />
199
  <edge from-layer="3" from-port="2" to-layer="4" to-port="2" />
200
  <edge from-layer="3" from-port="3" to-layer="4" to-port="3" />
201
- <edge from-layer="4" from-port="8" to-layer="5" to-port="4" />
202
  <edge from-layer="4" from-port="7" to-layer="5" to-port="3" />
 
203
  <edge from-layer="4" from-port="6" to-layer="5" to-port="2" />
204
  <edge from-layer="4" from-port="5" to-layer="5" to-port="1" />
205
  <edge from-layer="4" from-port="4" to-layer="5" to-port="0" />
206
- <edge from-layer="5" from-port="5" to-layer="8" to-port="0" />
207
- <edge from-layer="5" from-port="6" to-layer="8" to-port="1" />
208
- <edge from-layer="5" from-port="7" to-layer="8" to-port="2" />
209
  <edge from-layer="6" from-port="0" to-layer="8" to-port="3" />
210
  <edge from-layer="7" from-port="0" to-layer="8" to-port="4" />
211
  <edge from-layer="8" from-port="5" to-layer="9" to-port="0" />
@@ -215,5 +208,6 @@
215
  </edges>
216
  <rt_info>
217
  <eos_token_id value="50256" />
 
218
  </rt_info>
219
  </net>
 
1
  <?xml version="1.0"?>
2
  <net name="detokenizer" version="11">
3
  <layers>
4
+ <layer id="0" name="Parameter_62216" type="Parameter" version="opset1">
5
  <data shape="?,?" element_type="i64" />
6
  <output>
7
+ <port id="0" precision="I64" names="Parameter_62216">
8
  <dim>-1</dim>
9
  <dim>-1</dim>
10
  </port>
11
  </output>
12
  </layer>
13
+ <layer id="1" name="Convert_62235" type="Convert" version="opset1">
14
  <data destination_type="i32" />
15
  <input>
16
  <port id="0" precision="I64">
 
25
  </port>
26
  </output>
27
  </layer>
28
+ <layer id="2" name="Constant_62218" type="Const" version="opset1">
29
+ <data element_type="u8" shape="535153" offset="0" size="535153" />
30
  <output>
31
  <port id="0" precision="U8">
32
+ <dim>535153</dim>
33
  </port>
34
  </output>
35
  </layer>
36
+ <layer id="3" name="StringTensorUnpack_62219" type="StringTensorUnpack" version="extension">
37
  <data mode="begins_ends" />
38
  <input>
39
  <port id="0" precision="U8">
40
+ <dim>535153</dim>
41
  </port>
42
  </input>
43
  <output>
 
52
  </port>
53
  </output>
54
  </layer>
55
+ <layer id="4" name="VocabDecoder_62220" type="VocabDecoder" version="extension">
56
  <data skip_tokens="50256" />
57
  <input>
58
  <port id="0" precision="I32">
 
87
  </port>
88
  </output>
89
  </layer>
90
+ <layer id="5" name="FuzeRagged_62221" type="FuzeRagged" version="extension">
91
  <input>
92
  <port id="0" precision="I32">
93
  <dim>-1</dim>
 
101
  <port id="3" precision="I32">
102
  <dim>-1</dim>
103
  </port>
 
 
 
104
  </input>
105
  <output>
106
+ <port id="4" precision="I32">
 
 
 
107
  <dim>-1</dim>
108
  </port>
109
+ <port id="5" precision="I32">
110
  <dim>-1</dim>
111
  </port>
112
  </output>
113
  </layer>
114
+ <layer id="6" name="Constant_62223" type="Const" version="opset1">
115
+ <data element_type="u8" shape="47" offset="535153" size="47" />
116
  <output>
117
  <port id="0" precision="U8">
118
  <dim>47</dim>
119
  </port>
120
  </output>
121
  </layer>
122
+ <layer id="7" name="Constant_62225" type="Const" version="opset1">
123
+ <data element_type="u8" shape="2" offset="535200" size="2" />
124
  <output>
125
  <port id="0" precision="U8">
126
  <dim>2</dim>
127
  </port>
128
  </output>
129
  </layer>
130
+ <layer id="8" name="RegexNormalization_62226" type="RegexNormalization" version="extension">
131
  <data global_replace="true" />
132
  <input>
133
  <port id="0" precision="I32">
 
158
  </port>
159
  </output>
160
  </layer>
161
+ <layer id="9" name="StringTensorPack_62227" type="StringTensorPack" version="extension">
162
  <data mode="begins_ends" />
163
  <input>
164
  <port id="0" precision="I32">
 
177
  </port>
178
  </output>
179
  </layer>
180
+ <layer id="10" name="Result_62228" type="Result" version="opset1">
181
  <input>
182
  <port id="0" precision="STRING">
183
  <dim>-1</dim>
 
192
  <edge from-layer="3" from-port="1" to-layer="4" to-port="1" />
193
  <edge from-layer="3" from-port="2" to-layer="4" to-port="2" />
194
  <edge from-layer="3" from-port="3" to-layer="4" to-port="3" />
 
195
  <edge from-layer="4" from-port="7" to-layer="5" to-port="3" />
196
+ <edge from-layer="4" from-port="8" to-layer="8" to-port="2" />
197
  <edge from-layer="4" from-port="6" to-layer="5" to-port="2" />
198
  <edge from-layer="4" from-port="5" to-layer="5" to-port="1" />
199
  <edge from-layer="4" from-port="4" to-layer="5" to-port="0" />
200
+ <edge from-layer="5" from-port="4" to-layer="8" to-port="0" />
201
+ <edge from-layer="5" from-port="5" to-layer="8" to-port="1" />
 
202
  <edge from-layer="6" from-port="0" to-layer="8" to-port="3" />
203
  <edge from-layer="7" from-port="0" to-layer="8" to-port="4" />
204
  <edge from-layer="8" from-port="5" to-layer="9" to-port="0" />
 
208
  </edges>
209
  <rt_info>
210
  <eos_token_id value="50256" />
211
+ <original_tokenizer_class value="&lt;class 'transformers_modules.pytorch.tokenization_codegen25.CodeGen25Tokenizer'>" />
212
  </rt_info>
213
  </net>
openvino_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8677a6df83a63050a5eeb464d446e7d21dccc0b044d4ec6d0f3eb2d3da1e97f7
3
- size 6900888080
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e821b7ae930253385551533b2a000408a0930bcaa9888240f196f62eae6dde2c
3
+ size 6900888016
openvino_model.xml CHANGED
The diff for this file is too large to render. See raw diff
 
openvino_tokenizer.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7904174d95dfbed2580b9a957a5b21f14660d391f2f283b23b35c1a5f594559c
3
- size 1208600
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:348f1a1512e75e2213d7b2d2c2749c15e9458d56bae02b1336f7b502f881d509
3
+ size 1272940
openvino_tokenizer.xml CHANGED
@@ -1,21 +1,21 @@
1
  <?xml version="1.0"?>
2
  <net name="tokenizer" version="11">
3
  <layers>
4
- <layer id="0" name="Parameter_1" type="Parameter" version="opset1">
5
  <data shape="?" element_type="string" />
6
  <output>
7
- <port id="0" precision="STRING" names="Parameter_1">
8
  <dim>-1</dim>
9
  </port>
10
  </output>
11
  </layer>
12
- <layer id="1" name="Constant_8" type="Const" version="opset1">
13
  <data element_type="i64" shape="" offset="0" size="8" />
14
  <output>
15
  <port id="0" precision="I64" />
16
  </output>
17
  </layer>
18
- <layer id="2" name="StringTensorUnpack_2" type="StringTensorUnpack" version="extension">
19
  <data mode="begins_ends" />
20
  <input>
21
  <port id="0" precision="STRING">
@@ -34,32 +34,7 @@
34
  </port>
35
  </output>
36
  </layer>
37
- <layer id="3" name="NormalizeUnicode_3" type="NormalizeUnicode" version="extension">
38
- <data normalization_form="NFC" />
39
- <input>
40
- <port id="0" precision="I32">
41
- <dim>-1</dim>
42
- </port>
43
- <port id="1" precision="I32">
44
- <dim>-1</dim>
45
- </port>
46
- <port id="2" precision="U8">
47
- <dim>-1</dim>
48
- </port>
49
- </input>
50
- <output>
51
- <port id="3" precision="I32">
52
- <dim>-1</dim>
53
- </port>
54
- <port id="4" precision="I32">
55
- <dim>-1</dim>
56
- </port>
57
- <port id="5" precision="U8">
58
- <dim>-1</dim>
59
- </port>
60
- </output>
61
- </layer>
62
- <layer id="4" name="ShapeOf_4" type="ShapeOf" version="opset3">
63
  <data output_type="i64" />
64
  <input>
65
  <port id="0" precision="I32">
@@ -72,19 +47,19 @@
72
  </port>
73
  </output>
74
  </layer>
75
- <layer id="5" name="Constant_5" type="Const" version="opset1">
76
  <data element_type="i64" shape="" offset="0" size="8" />
77
  <output>
78
  <port id="0" precision="I64" />
79
  </output>
80
  </layer>
81
- <layer id="6" name="Constant_6" type="Const" version="opset1">
82
  <data element_type="i64" shape="" offset="0" size="8" />
83
  <output>
84
  <port id="0" precision="I64" />
85
  </output>
86
  </layer>
87
- <layer id="7" name="Gather_7" type="Gather" version="opset8">
88
  <data batch_dims="0" />
89
  <input>
90
  <port id="0" precision="I64">
@@ -97,13 +72,13 @@
97
  <port id="3" precision="I64" />
98
  </output>
99
  </layer>
100
- <layer id="8" name="Constant_9" type="Const" version="opset1">
101
  <data element_type="i64" shape="" offset="8" size="8" />
102
  <output>
103
  <port id="0" precision="I64" />
104
  </output>
105
  </layer>
106
- <layer id="9" name="Range_10" type="Range" version="opset4">
107
  <data output_type="i32" />
108
  <input>
109
  <port id="0" precision="I64" />
@@ -116,19 +91,19 @@
116
  </port>
117
  </output>
118
  </layer>
119
- <layer id="10" name="Constant_12" type="Const" version="opset1">
120
  <data element_type="i64" shape="" offset="8" size="8" />
121
  <output>
122
  <port id="0" precision="I64" />
123
  </output>
124
  </layer>
125
- <layer id="11" name="Constant_13" type="Const" version="opset1">
126
  <data element_type="i64" shape="" offset="8" size="8" />
127
  <output>
128
  <port id="0" precision="I64" />
129
  </output>
130
  </layer>
131
- <layer id="12" name="Add_14" type="Add" version="opset1">
132
  <data auto_broadcast="numpy" />
133
  <input>
134
  <port id="0" precision="I64" />
@@ -138,13 +113,13 @@
138
  <port id="2" precision="I64" />
139
  </output>
140
  </layer>
141
- <layer id="13" name="Constant_15" type="Const" version="opset1">
142
  <data element_type="i64" shape="" offset="8" size="8" />
143
  <output>
144
  <port id="0" precision="I64" />
145
  </output>
146
  </layer>
147
- <layer id="14" name="Range_16" type="Range" version="opset4">
148
  <data output_type="i32" />
149
  <input>
150
  <port id="0" precision="I64" />
@@ -157,16 +132,15 @@
157
  </port>
158
  </output>
159
  </layer>
160
- <layer id="15" name="Constant_79" type="Const" version="opset1">
161
- <data element_type="u8" shape="12817" offset="16" size="12817" />
162
  <output>
163
  <port id="0" precision="U8">
164
- <dim>12817</dim>
165
  </port>
166
  </output>
167
  </layer>
168
- <layer id="16" name="RegexSplit_80" type="RegexSplit" version="extension">
169
- <data behaviour="isolate" invert="false" max_splits="-1" />
170
  <input>
171
  <port id="0" precision="I32">
172
  <dim>-1</dim>
@@ -184,7 +158,7 @@
184
  <dim>-1</dim>
185
  </port>
186
  <port id="5" precision="U8">
187
- <dim>12817</dim>
188
  </port>
189
  </input>
190
  <output>
@@ -203,45 +177,13 @@
203
  <port id="10" precision="U8">
204
  <dim>-1</dim>
205
  </port>
206
- </output>
207
- </layer>
208
- <layer id="17" name="Constant_85" type="Const" version="opset1">
209
- <data element_type="u8" shape="79" offset="12833" size="79" />
210
- <output>
211
- <port id="0" precision="U8">
212
- <dim>79</dim>
213
- </port>
214
- </output>
215
- </layer>
216
- <layer id="18" name="Constant_82" type="Const" version="opset1">
217
- <data element_type="u8" shape="13303" offset="12912" size="13303" />
218
- <output>
219
- <port id="0" precision="U8">
220
- <dim>13303</dim>
221
- </port>
222
- </output>
223
- </layer>
224
- <layer id="19" name="StringTensorUnpack_83" type="StringTensorUnpack" version="extension">
225
- <data mode="begins_ends" />
226
- <input>
227
- <port id="0" precision="U8">
228
- <dim>13303</dim>
229
- </port>
230
- </input>
231
- <output>
232
- <port id="1" precision="I32">
233
- <dim>-1</dim>
234
- </port>
235
- <port id="2" precision="I32">
236
- <dim>-1</dim>
237
- </port>
238
- <port id="3" precision="U8">
239
  <dim>-1</dim>
240
  </port>
241
  </output>
242
  </layer>
243
- <layer id="20" name="RegexSplit_86" type="RegexSplit" version="extension">
244
- <data behaviour="contiguous" invert="false" max_splits="-1" />
245
  <input>
246
  <port id="0" precision="I32">
247
  <dim>-1</dim>
@@ -249,47 +191,38 @@
249
  <port id="1" precision="I32">
250
  <dim>-1</dim>
251
  </port>
252
- <port id="2" precision="I32">
253
- <dim>-1</dim>
254
- </port>
255
- <port id="3" precision="I32">
256
- <dim>-1</dim>
257
- </port>
258
- <port id="4" precision="U8">
259
- <dim>-1</dim>
260
- </port>
261
- <port id="5" precision="U8">
262
- <dim>79</dim>
263
- </port>
264
- <port id="6" precision="I32">
265
- <dim>-1</dim>
266
- </port>
267
- <port id="7" precision="I32">
268
  <dim>-1</dim>
269
  </port>
270
- <port id="8" precision="U8">
271
  <dim>-1</dim>
272
  </port>
273
  </input>
274
  <output>
275
- <port id="9" precision="I32">
276
  <dim>-1</dim>
277
  </port>
278
- <port id="10" precision="I32">
279
  <dim>-1</dim>
280
  </port>
281
- <port id="11" precision="I32">
282
  <dim>-1</dim>
283
  </port>
284
- <port id="12" precision="I32">
285
  <dim>-1</dim>
286
  </port>
287
- <port id="13" precision="U8">
288
- <dim>-1</dim>
 
 
 
 
 
289
  </port>
290
  </output>
291
  </layer>
292
- <layer id="21" name="BytesToChars_87" type="BytesToChars" version="extension">
 
293
  <input>
294
  <port id="0" precision="I32">
295
  <dim>-1</dim>
@@ -306,38 +239,47 @@
306
  <port id="4" precision="U8">
307
  <dim>-1</dim>
308
  </port>
 
 
 
 
 
 
309
  </input>
310
  <output>
311
- <port id="5" precision="I32">
312
  <dim>-1</dim>
313
  </port>
314
- <port id="6" precision="I32">
315
  <dim>-1</dim>
316
  </port>
317
- <port id="7" precision="I32">
318
  <dim>-1</dim>
319
  </port>
320
- <port id="8" precision="I32">
 
 
 
321
  <dim>-1</dim>
322
  </port>
323
- <port id="9" precision="U8">
324
  <dim>-1</dim>
325
  </port>
326
  </output>
327
  </layer>
328
- <layer id="22" name="Constant_89" type="Const" version="opset1">
329
- <data element_type="u8" shape="572279" offset="26215" size="572279" />
330
  <output>
331
  <port id="0" precision="U8">
332
- <dim>572279</dim>
333
  </port>
334
  </output>
335
  </layer>
336
- <layer id="23" name="StringTensorUnpack_90" type="StringTensorUnpack" version="extension">
337
  <data mode="begins_ends" />
338
  <input>
339
  <port id="0" precision="U8">
340
- <dim>572279</dim>
341
  </port>
342
  </input>
343
  <output>
@@ -352,19 +294,19 @@
352
  </port>
353
  </output>
354
  </layer>
355
- <layer id="24" name="Constant_170" type="Const" version="opset1">
356
- <data element_type="u8" shape="606330" offset="598494" size="606330" />
357
  <output>
358
  <port id="0" precision="U8">
359
- <dim>606330</dim>
360
  </port>
361
  </output>
362
  </layer>
363
- <layer id="25" name="StringTensorUnpack_171" type="StringTensorUnpack" version="extension">
364
  <data mode="begins_ends" />
365
  <input>
366
  <port id="0" precision="U8">
367
- <dim>606330</dim>
368
  </port>
369
  </input>
370
  <output>
@@ -379,25 +321,19 @@
379
  </port>
380
  </output>
381
  </layer>
382
- <layer id="26" name="Constant_98" type="Const" version="opset1">
383
- <data element_type="i64" shape="" offset="0" size="8" />
384
- <output>
385
- <port id="0" precision="I64" />
386
- </output>
387
- </layer>
388
- <layer id="27" name="Constant_92" type="Const" version="opset1">
389
- <data element_type="u8" shape="13303" offset="12912" size="13303" />
390
  <output>
391
  <port id="0" precision="U8">
392
- <dim>13303</dim>
393
  </port>
394
  </output>
395
  </layer>
396
- <layer id="28" name="StringTensorUnpack_93" type="StringTensorUnpack" version="extension">
397
  <data mode="begins_ends" />
398
  <input>
399
  <port id="0" precision="U8">
400
- <dim>13303</dim>
401
  </port>
402
  </input>
403
  <output>
@@ -412,150 +348,43 @@
412
  </port>
413
  </output>
414
  </layer>
415
- <layer id="29" name="ShapeOf_94" type="ShapeOf" version="opset3">
416
- <data output_type="i64" />
417
- <input>
418
- <port id="0" precision="I32">
419
- <dim>-1</dim>
420
- </port>
421
- </input>
422
  <output>
423
- <port id="1" precision="I64">
424
- <dim>1</dim>
425
- </port>
426
- </output>
427
- </layer>
428
- <layer id="30" name="Constant_95" type="Const" version="opset1">
429
- <data element_type="i64" shape="" offset="0" size="8" />
430
- <output>
431
- <port id="0" precision="I64" />
432
- </output>
433
- </layer>
434
- <layer id="31" name="Constant_96" type="Const" version="opset1">
435
- <data element_type="i64" shape="" offset="0" size="8" />
436
- <output>
437
- <port id="0" precision="I64" />
438
- </output>
439
- </layer>
440
- <layer id="32" name="Gather_97" type="Gather" version="opset8">
441
- <data batch_dims="0" />
442
- <input>
443
- <port id="0" precision="I64">
444
- <dim>1</dim>
445
  </port>
446
- <port id="1" precision="I64" />
447
- <port id="2" precision="I64" />
448
- </input>
449
- <output>
450
- <port id="3" precision="I64" />
451
- </output>
452
- </layer>
453
- <layer id="33" name="Constant_99" type="Const" version="opset1">
454
- <data element_type="i64" shape="" offset="8" size="8" />
455
- <output>
456
- <port id="0" precision="I64" />
457
  </output>
458
  </layer>
459
- <layer id="34" name="Range_100" type="Range" version="opset4">
460
- <data output_type="i32" />
461
  <input>
462
- <port id="0" precision="I64" />
463
- <port id="1" precision="I64" />
464
- <port id="2" precision="I64" />
465
- </input>
466
- <output>
467
- <port id="3" precision="I32">
468
- <dim>-1</dim>
469
  </port>
470
- </output>
471
- </layer>
472
- <layer id="35" name="Constant_102" type="Const" version="opset1">
473
- <data element_type="i64" shape="" offset="8" size="8" />
474
- <output>
475
- <port id="0" precision="I64" />
476
- </output>
477
- </layer>
478
- <layer id="36" name="Constant_103" type="Const" version="opset1">
479
- <data element_type="i64" shape="" offset="8" size="8" />
480
- <output>
481
- <port id="0" precision="I64" />
482
- </output>
483
- </layer>
484
- <layer id="37" name="Add_104" type="Add" version="opset1">
485
- <data auto_broadcast="numpy" />
486
- <input>
487
- <port id="0" precision="I64" />
488
- <port id="1" precision="I64" />
489
- </input>
490
- <output>
491
- <port id="2" precision="I64" />
492
- </output>
493
- </layer>
494
- <layer id="38" name="Constant_105" type="Const" version="opset1">
495
- <data element_type="i64" shape="" offset="8" size="8" />
496
- <output>
497
- <port id="0" precision="I64" />
498
- </output>
499
- </layer>
500
- <layer id="39" name="Range_106" type="Range" version="opset4">
501
- <data output_type="i32" />
502
- <input>
503
- <port id="0" precision="I64" />
504
- <port id="1" precision="I64" />
505
- <port id="2" precision="I64" />
506
  </input>
507
  <output>
508
- <port id="3" precision="I32">
509
- <dim>-1</dim>
510
- </port>
511
- </output>
512
- </layer>
513
- <layer id="40" name="BytesToChars_168" type="BytesToChars" version="extension">
514
- <input>
515
- <port id="0" precision="I32">
516
- <dim>-1</dim>
517
- </port>
518
  <port id="1" precision="I32">
519
  <dim>-1</dim>
520
  </port>
521
  <port id="2" precision="I32">
522
  <dim>-1</dim>
523
  </port>
524
- <port id="3" precision="I32">
525
- <dim>-1</dim>
526
- </port>
527
- <port id="4" precision="U8">
528
- <dim>-1</dim>
529
- </port>
530
- </input>
531
- <output>
532
- <port id="5" precision="I32">
533
- <dim>-1</dim>
534
- </port>
535
- <port id="6" precision="I32">
536
- <dim>-1</dim>
537
- </port>
538
- <port id="7" precision="I32">
539
- <dim>-1</dim>
540
- </port>
541
- <port id="8" precision="I32">
542
- <dim>-1</dim>
543
- </port>
544
- <port id="9" precision="U8">
545
  <dim>-1</dim>
546
  </port>
547
  </output>
548
  </layer>
549
- <layer id="41" name="Constant_172" type="Const" version="opset1">
550
- <data element_type="i32" shape="942" offset="1204824" size="3768" />
551
  <output>
552
  <port id="0" precision="I32">
553
  <dim>942</dim>
554
  </port>
555
  </output>
556
  </layer>
557
- <layer id="42" name="BPETokenizer_173" type="BPETokenizer" version="extension">
558
- <data unk_token="" fuse_unk="false" suffix_indicator="" end_suffix="" byte_fallback="false" />
559
  <input>
560
  <port id="0" precision="I32">
561
  <dim>-1</dim>
@@ -600,22 +429,31 @@
600
  <dim>-1</dim>
601
  </port>
602
  <port id="14" precision="I32">
 
 
 
 
 
 
 
 
 
603
  <dim>942</dim>
604
  </port>
605
  </input>
606
  <output>
607
- <port id="15" precision="I32">
608
  <dim>-1</dim>
609
  </port>
610
- <port id="16" precision="I32">
611
  <dim>-1</dim>
612
  </port>
613
- <port id="17" precision="I32">
614
  <dim>-1</dim>
615
  </port>
616
  </output>
617
  </layer>
618
- <layer id="43" name="Subtract_174" type="Subtract" version="opset1">
619
  <data auto_broadcast="numpy" />
620
  <input>
621
  <port id="0" precision="I32">
@@ -631,13 +469,13 @@
631
  </port>
632
  </output>
633
  </layer>
634
- <layer id="44" name="Constant_175" type="Const" version="opset1">
635
- <data element_type="i32" shape="" offset="1208592" size="4" />
636
  <output>
637
  <port id="0" precision="I32" />
638
  </output>
639
  </layer>
640
- <layer id="45" name="Minimum_176" type="Minimum" version="opset1">
641
  <data auto_broadcast="numpy" />
642
  <input>
643
  <port id="0" precision="I32">
@@ -651,7 +489,7 @@
651
  </port>
652
  </output>
653
  </layer>
654
- <layer id="46" name="Add_177" type="Add" version="opset1">
655
  <data auto_broadcast="numpy" />
656
  <input>
657
  <port id="0" precision="I32">
@@ -667,7 +505,7 @@
667
  </port>
668
  </output>
669
  </layer>
670
- <layer id="47" name="Subtract_178" type="Subtract" version="opset1">
671
  <data auto_broadcast="numpy" />
672
  <input>
673
  <port id="0" precision="I32">
@@ -683,13 +521,13 @@
683
  </port>
684
  </output>
685
  </layer>
686
- <layer id="48" name="Constant_179" type="Const" version="opset1">
687
- <data element_type="i32" shape="" offset="1208596" size="4" />
688
  <output>
689
  <port id="0" precision="I32" />
690
  </output>
691
  </layer>
692
- <layer id="49" name="ReduceMax_180" type="ReduceMax" version="opset1">
693
  <data keep_dims="false" />
694
  <input>
695
  <port id="0" precision="I32">
@@ -701,14 +539,14 @@
701
  <port id="2" precision="I32" />
702
  </output>
703
  </layer>
704
- <layer id="50" name="Constant_181" type="Const" version="opset1">
705
- <data element_type="i32" shape="" offset="1208596" size="4" />
706
  <output>
707
  <port id="0" precision="I32" />
708
  </output>
709
  </layer>
710
- <layer id="51" name="RaggedToDense_182" type="RaggedToDense" version="extension">
711
- <data pad_right="true" />
712
  <input>
713
  <port id="0" precision="I32">
714
  <dim>-1</dim>
@@ -733,7 +571,7 @@
733
  </port>
734
  </output>
735
  </layer>
736
- <layer id="52" name="Convert_183" type="Convert" version="opset1">
737
  <data destination_type="i32" />
738
  <input>
739
  <port id="0" precision="BOOL">
@@ -748,7 +586,7 @@
748
  </port>
749
  </output>
750
  </layer>
751
- <layer id="53" name="Convert_183" type="Convert" version="opset1">
752
  <data destination_type="i64" />
753
  <input>
754
  <port id="0" precision="I32">
@@ -763,7 +601,7 @@
763
  </port>
764
  </output>
765
  </layer>
766
- <layer id="55" name="RaggedToDense_182.0" type="Convert" version="opset1">
767
  <data destination_type="i64" />
768
  <input>
769
  <port id="0" precision="I32">
@@ -778,7 +616,7 @@
778
  </port>
779
  </output>
780
  </layer>
781
- <layer id="56" name="Result_184" type="Result" version="opset1">
782
  <input>
783
  <port id="0" precision="I64">
784
  <dim>-1</dim>
@@ -786,7 +624,7 @@
786
  </port>
787
  </input>
788
  </layer>
789
- <layer id="54" name="Result_185" type="Result" version="opset1">
790
  <input>
791
  <port id="0" precision="I64">
792
  <dim>-1</dim>
@@ -797,99 +635,80 @@
797
  </layers>
798
  <edges>
799
  <edge from-layer="0" from-port="0" to-layer="2" to-port="0" />
800
- <edge from-layer="1" from-port="0" to-layer="9" to-port="0" />
801
  <edge from-layer="2" from-port="1" to-layer="3" to-port="0" />
802
- <edge from-layer="2" from-port="2" to-layer="3" to-port="1" />
803
- <edge from-layer="2" from-port="3" to-layer="3" to-port="2" />
804
- <edge from-layer="3" from-port="3" to-layer="4" to-port="0" />
805
- <edge from-layer="3" from-port="5" to-layer="16" to-port="4" />
806
- <edge from-layer="3" from-port="4" to-layer="16" to-port="3" />
807
- <edge from-layer="3" from-port="3" to-layer="16" to-port="2" />
808
- <edge from-layer="4" from-port="1" to-layer="7" to-port="0" />
809
- <edge from-layer="5" from-port="0" to-layer="7" to-port="1" />
810
- <edge from-layer="6" from-port="0" to-layer="7" to-port="2" />
811
- <edge from-layer="7" from-port="3" to-layer="9" to-port="1" />
812
- <edge from-layer="7" from-port="3" to-layer="12" to-port="0" />
813
- <edge from-layer="8" from-port="0" to-layer="9" to-port="2" />
814
- <edge from-layer="9" from-port="3" to-layer="16" to-port="0" />
815
- <edge from-layer="10" from-port="0" to-layer="14" to-port="0" />
816
- <edge from-layer="11" from-port="0" to-layer="12" to-port="1" />
817
- <edge from-layer="12" from-port="2" to-layer="14" to-port="1" />
818
- <edge from-layer="13" from-port="0" to-layer="14" to-port="2" />
819
- <edge from-layer="14" from-port="3" to-layer="16" to-port="1" />
820
- <edge from-layer="15" from-port="0" to-layer="16" to-port="5" />
821
- <edge from-layer="16" from-port="6" to-layer="20" to-port="0" />
822
- <edge from-layer="16" from-port="7" to-layer="20" to-port="1" />
823
- <edge from-layer="16" from-port="8" to-layer="20" to-port="2" />
824
- <edge from-layer="16" from-port="9" to-layer="20" to-port="3" />
825
- <edge from-layer="16" from-port="10" to-layer="20" to-port="4" />
826
- <edge from-layer="17" from-port="0" to-layer="20" to-port="5" />
827
- <edge from-layer="18" from-port="0" to-layer="19" to-port="0" />
828
- <edge from-layer="19" from-port="3" to-layer="20" to-port="8" />
829
- <edge from-layer="19" from-port="2" to-layer="20" to-port="7" />
830
- <edge from-layer="19" from-port="1" to-layer="20" to-port="6" />
831
- <edge from-layer="20" from-port="9" to-layer="21" to-port="0" />
832
- <edge from-layer="20" from-port="10" to-layer="21" to-port="1" />
833
- <edge from-layer="20" from-port="11" to-layer="21" to-port="2" />
834
- <edge from-layer="20" from-port="12" to-layer="21" to-port="3" />
835
- <edge from-layer="20" from-port="13" to-layer="21" to-port="4" />
836
- <edge from-layer="21" from-port="9" to-layer="42" to-port="4" />
837
- <edge from-layer="21" from-port="8" to-layer="42" to-port="3" />
838
- <edge from-layer="21" from-port="7" to-layer="42" to-port="2" />
839
- <edge from-layer="21" from-port="6" to-layer="42" to-port="1" />
840
- <edge from-layer="21" from-port="5" to-layer="42" to-port="0" />
841
- <edge from-layer="22" from-port="0" to-layer="23" to-port="0" />
842
- <edge from-layer="23" from-port="1" to-layer="42" to-port="5" />
843
- <edge from-layer="23" from-port="2" to-layer="42" to-port="6" />
844
- <edge from-layer="23" from-port="3" to-layer="42" to-port="7" />
845
- <edge from-layer="24" from-port="0" to-layer="25" to-port="0" />
846
- <edge from-layer="25" from-port="3" to-layer="42" to-port="10" />
847
- <edge from-layer="25" from-port="2" to-layer="42" to-port="9" />
848
- <edge from-layer="25" from-port="1" to-layer="42" to-port="8" />
849
- <edge from-layer="26" from-port="0" to-layer="34" to-port="0" />
850
- <edge from-layer="27" from-port="0" to-layer="28" to-port="0" />
851
- <edge from-layer="28" from-port="1" to-layer="40" to-port="2" />
852
- <edge from-layer="28" from-port="2" to-layer="40" to-port="3" />
853
- <edge from-layer="28" from-port="3" to-layer="40" to-port="4" />
854
- <edge from-layer="28" from-port="1" to-layer="29" to-port="0" />
855
- <edge from-layer="29" from-port="1" to-layer="32" to-port="0" />
856
- <edge from-layer="30" from-port="0" to-layer="32" to-port="1" />
857
- <edge from-layer="31" from-port="0" to-layer="32" to-port="2" />
858
- <edge from-layer="32" from-port="3" to-layer="37" to-port="0" />
859
- <edge from-layer="32" from-port="3" to-layer="34" to-port="1" />
860
- <edge from-layer="33" from-port="0" to-layer="34" to-port="2" />
861
- <edge from-layer="34" from-port="3" to-layer="40" to-port="0" />
862
- <edge from-layer="35" from-port="0" to-layer="39" to-port="0" />
863
- <edge from-layer="36" from-port="0" to-layer="37" to-port="1" />
864
- <edge from-layer="37" from-port="2" to-layer="39" to-port="1" />
865
- <edge from-layer="38" from-port="0" to-layer="39" to-port="2" />
866
- <edge from-layer="39" from-port="3" to-layer="40" to-port="1" />
867
- <edge from-layer="40" from-port="7" to-layer="42" to-port="11" />
868
- <edge from-layer="40" from-port="8" to-layer="42" to-port="12" />
869
- <edge from-layer="40" from-port="9" to-layer="42" to-port="13" />
870
- <edge from-layer="41" from-port="0" to-layer="42" to-port="14" />
871
- <edge from-layer="42" from-port="16" to-layer="43" to-port="0" />
872
- <edge from-layer="42" from-port="15" to-layer="43" to-port="1" />
873
- <edge from-layer="42" from-port="15" to-layer="46" to-port="0" />
874
- <edge from-layer="42" from-port="17" to-layer="51" to-port="2" />
875
- <edge from-layer="42" from-port="15" to-layer="47" to-port="1" />
876
- <edge from-layer="42" from-port="15" to-layer="51" to-port="0" />
877
- <edge from-layer="43" from-port="2" to-layer="45" to-port="0" />
878
- <edge from-layer="44" from-port="0" to-layer="45" to-port="1" />
879
- <edge from-layer="45" from-port="2" to-layer="46" to-port="1" />
880
- <edge from-layer="46" from-port="2" to-layer="51" to-port="1" />
881
- <edge from-layer="46" from-port="2" to-layer="47" to-port="0" />
882
- <edge from-layer="47" from-port="2" to-layer="49" to-port="0" />
883
- <edge from-layer="48" from-port="0" to-layer="49" to-port="1" />
884
- <edge from-layer="49" from-port="2" to-layer="51" to-port="3" />
885
- <edge from-layer="50" from-port="0" to-layer="51" to-port="4" />
886
- <edge from-layer="51" from-port="6" to-layer="52" to-port="0" />
887
- <edge from-layer="51" from-port="5" to-layer="55" to-port="0" />
888
- <edge from-layer="52" from-port="1" to-layer="53" to-port="0" />
889
- <edge from-layer="53" from-port="1" to-layer="54" to-port="0" />
890
- <edge from-layer="55" from-port="1" to-layer="56" to-port="0" />
891
  </edges>
892
  <rt_info>
893
  <eos_token_id value="50256" />
 
894
  </rt_info>
895
  </net>
 
1
  <?xml version="1.0"?>
2
  <net name="tokenizer" version="11">
3
  <layers>
4
+ <layer id="0" name="Parameter_62109" type="Parameter" version="opset1">
5
  <data shape="?" element_type="string" />
6
  <output>
7
+ <port id="0" precision="STRING" names="Parameter_62109">
8
  <dim>-1</dim>
9
  </port>
10
  </output>
11
  </layer>
12
+ <layer id="1" name="Constant_62115" type="Const" version="opset1">
13
  <data element_type="i64" shape="" offset="0" size="8" />
14
  <output>
15
  <port id="0" precision="I64" />
16
  </output>
17
  </layer>
18
+ <layer id="2" name="StringTensorUnpack_62110" type="StringTensorUnpack" version="extension">
19
  <data mode="begins_ends" />
20
  <input>
21
  <port id="0" precision="STRING">
 
34
  </port>
35
  </output>
36
  </layer>
37
+ <layer id="3" name="ShapeOf_62111" type="ShapeOf" version="opset3">
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
38
  <data output_type="i64" />
39
  <input>
40
  <port id="0" precision="I32">
 
47
  </port>
48
  </output>
49
  </layer>
50
+ <layer id="4" name="Constant_62112" type="Const" version="opset1">
51
  <data element_type="i64" shape="" offset="0" size="8" />
52
  <output>
53
  <port id="0" precision="I64" />
54
  </output>
55
  </layer>
56
+ <layer id="5" name="Constant_62113" type="Const" version="opset1">
57
  <data element_type="i64" shape="" offset="0" size="8" />
58
  <output>
59
  <port id="0" precision="I64" />
60
  </output>
61
  </layer>
62
+ <layer id="6" name="Gather_62114" type="Gather" version="opset8">
63
  <data batch_dims="0" />
64
  <input>
65
  <port id="0" precision="I64">
 
72
  <port id="3" precision="I64" />
73
  </output>
74
  </layer>
75
+ <layer id="7" name="Constant_62116" type="Const" version="opset1">
76
  <data element_type="i64" shape="" offset="8" size="8" />
77
  <output>
78
  <port id="0" precision="I64" />
79
  </output>
80
  </layer>
81
+ <layer id="8" name="Range_62117" type="Range" version="opset4">
82
  <data output_type="i32" />
83
  <input>
84
  <port id="0" precision="I64" />
 
91
  </port>
92
  </output>
93
  </layer>
94
+ <layer id="9" name="Constant_62118" type="Const" version="opset1">
95
  <data element_type="i64" shape="" offset="8" size="8" />
96
  <output>
97
  <port id="0" precision="I64" />
98
  </output>
99
  </layer>
100
+ <layer id="10" name="Constant_62119" type="Const" version="opset1">
101
  <data element_type="i64" shape="" offset="8" size="8" />
102
  <output>
103
  <port id="0" precision="I64" />
104
  </output>
105
  </layer>
106
+ <layer id="11" name="Add_62120" type="Add" version="opset1">
107
  <data auto_broadcast="numpy" />
108
  <input>
109
  <port id="0" precision="I64" />
 
113
  <port id="2" precision="I64" />
114
  </output>
115
  </layer>
116
+ <layer id="12" name="Constant_62121" type="Const" version="opset1">
117
  <data element_type="i64" shape="" offset="8" size="8" />
118
  <output>
119
  <port id="0" precision="I64" />
120
  </output>
121
  </layer>
122
+ <layer id="13" name="Range_62122" type="Range" version="opset4">
123
  <data output_type="i32" />
124
  <input>
125
  <port id="0" precision="I64" />
 
132
  </port>
133
  </output>
134
  </layer>
135
+ <layer id="14" name="Constant_62184" type="Const" version="opset1">
136
+ <data element_type="u8" shape="19" offset="16" size="19" />
137
  <output>
138
  <port id="0" precision="U8">
139
+ <dim>19</dim>
140
  </port>
141
  </output>
142
  </layer>
143
+ <layer id="15" name="SpecialTokensSplit_62185" type="SpecialTokensSplit" version="extension">
 
144
  <input>
145
  <port id="0" precision="I32">
146
  <dim>-1</dim>
 
158
  <dim>-1</dim>
159
  </port>
160
  <port id="5" precision="U8">
161
+ <dim>19</dim>
162
  </port>
163
  </input>
164
  <output>
 
177
  <port id="10" precision="U8">
178
  <dim>-1</dim>
179
  </port>
180
+ <port id="11" precision="BOOL">
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
181
  <dim>-1</dim>
182
  </port>
183
  </output>
184
  </layer>
185
+ <layer id="16" name="NormalizeUnicode_62186" type="NormalizeUnicode" version="extension">
186
+ <data normalization_form="NFC" />
187
  <input>
188
  <port id="0" precision="I32">
189
  <dim>-1</dim>
 
191
  <port id="1" precision="I32">
192
  <dim>-1</dim>
193
  </port>
194
+ <port id="2" precision="U8">
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
195
  <dim>-1</dim>
196
  </port>
197
+ <port id="3" precision="BOOL">
198
  <dim>-1</dim>
199
  </port>
200
  </input>
201
  <output>
202
+ <port id="4" precision="I32">
203
  <dim>-1</dim>
204
  </port>
205
+ <port id="5" precision="I32">
206
  <dim>-1</dim>
207
  </port>
208
+ <port id="6" precision="U8">
209
  <dim>-1</dim>
210
  </port>
211
+ <port id="7" precision="BOOL">
212
  <dim>-1</dim>
213
  </port>
214
+ </output>
215
+ </layer>
216
+ <layer id="17" name="Constant_62188" type="Const" version="opset1">
217
+ <data element_type="u8" shape="79" offset="35" size="79" />
218
+ <output>
219
+ <port id="0" precision="U8">
220
+ <dim>79</dim>
221
  </port>
222
  </output>
223
  </layer>
224
+ <layer id="18" name="RegexSplit_62189" type="RegexSplit" version="extension">
225
+ <data behaviour="contiguous" invert="false" max_splits="-1" />
226
  <input>
227
  <port id="0" precision="I32">
228
  <dim>-1</dim>
 
239
  <port id="4" precision="U8">
240
  <dim>-1</dim>
241
  </port>
242
+ <port id="5" precision="BOOL">
243
+ <dim>-1</dim>
244
+ </port>
245
+ <port id="6" precision="U8">
246
+ <dim>79</dim>
247
+ </port>
248
  </input>
249
  <output>
250
+ <port id="7" precision="I32">
251
  <dim>-1</dim>
252
  </port>
253
+ <port id="8" precision="I32">
254
  <dim>-1</dim>
255
  </port>
256
+ <port id="9" precision="I32">
257
  <dim>-1</dim>
258
  </port>
259
+ <port id="10" precision="I32">
260
+ <dim>-1</dim>
261
+ </port>
262
+ <port id="11" precision="U8">
263
  <dim>-1</dim>
264
  </port>
265
+ <port id="12" precision="BOOL">
266
  <dim>-1</dim>
267
  </port>
268
  </output>
269
  </layer>
270
+ <layer id="19" name="Constant_62191" type="Const" version="opset1">
271
+ <data element_type="u8" shape="535153" offset="114" size="535153" />
272
  <output>
273
  <port id="0" precision="U8">
274
+ <dim>535153</dim>
275
  </port>
276
  </output>
277
  </layer>
278
+ <layer id="20" name="StringTensorUnpack_62192" type="StringTensorUnpack" version="extension">
279
  <data mode="begins_ends" />
280
  <input>
281
  <port id="0" precision="U8">
282
+ <dim>535153</dim>
283
  </port>
284
  </input>
285
  <output>
 
294
  </port>
295
  </output>
296
  </layer>
297
+ <layer id="21" name="Constant_62197" type="Const" version="opset1">
298
+ <data element_type="u8" shape="367533" offset="535267" size="367533" />
299
  <output>
300
  <port id="0" precision="U8">
301
+ <dim>367533</dim>
302
  </port>
303
  </output>
304
  </layer>
305
+ <layer id="22" name="StringTensorUnpack_62198" type="StringTensorUnpack" version="extension">
306
  <data mode="begins_ends" />
307
  <input>
308
  <port id="0" precision="U8">
309
+ <dim>367533</dim>
310
  </port>
311
  </input>
312
  <output>
 
321
  </port>
322
  </output>
323
  </layer>
324
+ <layer id="23" name="Constant_62200" type="Const" version="opset1">
325
+ <data element_type="u8" shape="353061" offset="902800" size="353061" />
 
 
 
 
 
 
326
  <output>
327
  <port id="0" precision="U8">
328
+ <dim>353061</dim>
329
  </port>
330
  </output>
331
  </layer>
332
+ <layer id="24" name="StringTensorUnpack_62201" type="StringTensorUnpack" version="extension">
333
  <data mode="begins_ends" />
334
  <input>
335
  <port id="0" precision="U8">
336
+ <dim>353061</dim>
337
  </port>
338
  </input>
339
  <output>
 
348
  </port>
349
  </output>
350
  </layer>
351
+ <layer id="25" name="Constant_62194" type="Const" version="opset1">
352
+ <data element_type="u8" shape="13303" offset="1255861" size="13303" />
 
 
 
 
 
353
  <output>
354
+ <port id="0" precision="U8">
355
+ <dim>13303</dim>
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
356
  </port>
 
 
 
 
 
 
 
 
 
 
 
357
  </output>
358
  </layer>
359
+ <layer id="26" name="StringTensorUnpack_62195" type="StringTensorUnpack" version="extension">
360
+ <data mode="begins_ends" />
361
  <input>
362
+ <port id="0" precision="U8">
363
+ <dim>13303</dim>
 
 
 
 
 
364
  </port>
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
365
  </input>
366
  <output>
 
 
 
 
 
 
 
 
 
 
367
  <port id="1" precision="I32">
368
  <dim>-1</dim>
369
  </port>
370
  <port id="2" precision="I32">
371
  <dim>-1</dim>
372
  </port>
373
+ <port id="3" precision="U8">
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
374
  <dim>-1</dim>
375
  </port>
376
  </output>
377
  </layer>
378
+ <layer id="27" name="Constant_62202" type="Const" version="opset1">
379
+ <data element_type="i32" shape="942" offset="1269164" size="3768" />
380
  <output>
381
  <port id="0" precision="I32">
382
  <dim>942</dim>
383
  </port>
384
  </output>
385
  </layer>
386
+ <layer id="28" name="BPETokenizer_62203" type="BPETokenizer" version="extension">
387
+ <data unk_token="" fuse_unk="false" suffix_indicator="" end_suffix="" byte_fallback="false" cache_capacity="20000" />
388
  <input>
389
  <port id="0" precision="I32">
390
  <dim>-1</dim>
 
429
  <dim>-1</dim>
430
  </port>
431
  <port id="14" precision="I32">
432
+ <dim>-1</dim>
433
+ </port>
434
+ <port id="15" precision="I32">
435
+ <dim>-1</dim>
436
+ </port>
437
+ <port id="16" precision="U8">
438
+ <dim>-1</dim>
439
+ </port>
440
+ <port id="17" precision="I32">
441
  <dim>942</dim>
442
  </port>
443
  </input>
444
  <output>
445
+ <port id="18" precision="I32">
446
  <dim>-1</dim>
447
  </port>
448
+ <port id="19" precision="I32">
449
  <dim>-1</dim>
450
  </port>
451
+ <port id="20" precision="I32">
452
  <dim>-1</dim>
453
  </port>
454
  </output>
455
  </layer>
456
+ <layer id="29" name="Subtract_62204" type="Subtract" version="opset1">
457
  <data auto_broadcast="numpy" />
458
  <input>
459
  <port id="0" precision="I32">
 
469
  </port>
470
  </output>
471
  </layer>
472
+ <layer id="30" name="Constant_62205" type="Const" version="opset1">
473
+ <data element_type="i32" shape="" offset="1272932" size="4" />
474
  <output>
475
  <port id="0" precision="I32" />
476
  </output>
477
  </layer>
478
+ <layer id="31" name="Minimum_62206" type="Minimum" version="opset1">
479
  <data auto_broadcast="numpy" />
480
  <input>
481
  <port id="0" precision="I32">
 
489
  </port>
490
  </output>
491
  </layer>
492
+ <layer id="32" name="Subtract_62207" type="Subtract" version="opset1">
493
  <data auto_broadcast="numpy" />
494
  <input>
495
  <port id="0" precision="I32">
 
505
  </port>
506
  </output>
507
  </layer>
508
+ <layer id="33" name="Subtract_62208" type="Subtract" version="opset1">
509
  <data auto_broadcast="numpy" />
510
  <input>
511
  <port id="0" precision="I32">
 
521
  </port>
522
  </output>
523
  </layer>
524
+ <layer id="34" name="Constant_62209" type="Const" version="opset1">
525
+ <data element_type="i32" shape="" offset="1272936" size="4" />
526
  <output>
527
  <port id="0" precision="I32" />
528
  </output>
529
  </layer>
530
+ <layer id="35" name="ReduceMax_62210" type="ReduceMax" version="opset1">
531
  <data keep_dims="false" />
532
  <input>
533
  <port id="0" precision="I32">
 
539
  <port id="2" precision="I32" />
540
  </output>
541
  </layer>
542
+ <layer id="36" name="Constant_62211" type="Const" version="opset1">
543
+ <data element_type="i32" shape="" offset="1272936" size="4" />
544
  <output>
545
  <port id="0" precision="I32" />
546
  </output>
547
  </layer>
548
+ <layer id="37" name="RaggedToDense_62212" type="RaggedToDense" version="extension">
549
+ <data pad_right="false" />
550
  <input>
551
  <port id="0" precision="I32">
552
  <dim>-1</dim>
 
571
  </port>
572
  </output>
573
  </layer>
574
+ <layer id="38" name="Convert_62213" type="Convert" version="opset1">
575
  <data destination_type="i32" />
576
  <input>
577
  <port id="0" precision="BOOL">
 
586
  </port>
587
  </output>
588
  </layer>
589
+ <layer id="39" name="Convert_62213" type="Convert" version="opset1">
590
  <data destination_type="i64" />
591
  <input>
592
  <port id="0" precision="I32">
 
601
  </port>
602
  </output>
603
  </layer>
604
+ <layer id="41" name="RaggedToDense_62212.0" type="Convert" version="opset1">
605
  <data destination_type="i64" />
606
  <input>
607
  <port id="0" precision="I32">
 
616
  </port>
617
  </output>
618
  </layer>
619
+ <layer id="42" name="Result_62214" type="Result" version="opset1">
620
  <input>
621
  <port id="0" precision="I64">
622
  <dim>-1</dim>
 
624
  </port>
625
  </input>
626
  </layer>
627
+ <layer id="40" name="Result_62215" type="Result" version="opset1">
628
  <input>
629
  <port id="0" precision="I64">
630
  <dim>-1</dim>
 
635
  </layers>
636
  <edges>
637
  <edge from-layer="0" from-port="0" to-layer="2" to-port="0" />
638
+ <edge from-layer="1" from-port="0" to-layer="8" to-port="0" />
639
  <edge from-layer="2" from-port="1" to-layer="3" to-port="0" />
640
+ <edge from-layer="2" from-port="3" to-layer="15" to-port="4" />
641
+ <edge from-layer="2" from-port="2" to-layer="15" to-port="3" />
642
+ <edge from-layer="2" from-port="1" to-layer="15" to-port="2" />
643
+ <edge from-layer="3" from-port="1" to-layer="6" to-port="0" />
644
+ <edge from-layer="4" from-port="0" to-layer="6" to-port="1" />
645
+ <edge from-layer="5" from-port="0" to-layer="6" to-port="2" />
646
+ <edge from-layer="6" from-port="3" to-layer="11" to-port="0" />
647
+ <edge from-layer="6" from-port="3" to-layer="8" to-port="1" />
648
+ <edge from-layer="7" from-port="0" to-layer="8" to-port="2" />
649
+ <edge from-layer="8" from-port="3" to-layer="15" to-port="0" />
650
+ <edge from-layer="9" from-port="0" to-layer="13" to-port="0" />
651
+ <edge from-layer="10" from-port="0" to-layer="11" to-port="1" />
652
+ <edge from-layer="11" from-port="2" to-layer="13" to-port="1" />
653
+ <edge from-layer="12" from-port="0" to-layer="13" to-port="2" />
654
+ <edge from-layer="13" from-port="3" to-layer="15" to-port="1" />
655
+ <edge from-layer="14" from-port="0" to-layer="15" to-port="5" />
656
+ <edge from-layer="15" from-port="9" to-layer="16" to-port="1" />
657
+ <edge from-layer="15" from-port="7" to-layer="18" to-port="1" />
658
+ <edge from-layer="15" from-port="6" to-layer="18" to-port="0" />
659
+ <edge from-layer="15" from-port="11" to-layer="16" to-port="3" />
660
+ <edge from-layer="15" from-port="10" to-layer="16" to-port="2" />
661
+ <edge from-layer="15" from-port="8" to-layer="16" to-port="0" />
662
+ <edge from-layer="16" from-port="4" to-layer="18" to-port="2" />
663
+ <edge from-layer="16" from-port="5" to-layer="18" to-port="3" />
664
+ <edge from-layer="16" from-port="6" to-layer="18" to-port="4" />
665
+ <edge from-layer="16" from-port="7" to-layer="18" to-port="5" />
666
+ <edge from-layer="17" from-port="0" to-layer="18" to-port="6" />
667
+ <edge from-layer="18" from-port="11" to-layer="28" to-port="4" />
668
+ <edge from-layer="18" from-port="10" to-layer="28" to-port="3" />
669
+ <edge from-layer="18" from-port="9" to-layer="28" to-port="2" />
670
+ <edge from-layer="18" from-port="8" to-layer="28" to-port="1" />
671
+ <edge from-layer="18" from-port="7" to-layer="28" to-port="0" />
672
+ <edge from-layer="19" from-port="0" to-layer="20" to-port="0" />
673
+ <edge from-layer="20" from-port="1" to-layer="28" to-port="5" />
674
+ <edge from-layer="20" from-port="2" to-layer="28" to-port="6" />
675
+ <edge from-layer="20" from-port="3" to-layer="28" to-port="7" />
676
+ <edge from-layer="21" from-port="0" to-layer="22" to-port="0" />
677
+ <edge from-layer="22" from-port="1" to-layer="28" to-port="8" />
678
+ <edge from-layer="22" from-port="2" to-layer="28" to-port="9" />
679
+ <edge from-layer="22" from-port="3" to-layer="28" to-port="10" />
680
+ <edge from-layer="23" from-port="0" to-layer="24" to-port="0" />
681
+ <edge from-layer="24" from-port="1" to-layer="28" to-port="11" />
682
+ <edge from-layer="24" from-port="2" to-layer="28" to-port="12" />
683
+ <edge from-layer="24" from-port="3" to-layer="28" to-port="13" />
684
+ <edge from-layer="25" from-port="0" to-layer="26" to-port="0" />
685
+ <edge from-layer="26" from-port="1" to-layer="28" to-port="14" />
686
+ <edge from-layer="26" from-port="2" to-layer="28" to-port="15" />
687
+ <edge from-layer="26" from-port="3" to-layer="28" to-port="16" />
688
+ <edge from-layer="27" from-port="0" to-layer="28" to-port="17" />
689
+ <edge from-layer="28" from-port="20" to-layer="37" to-port="2" />
690
+ <edge from-layer="28" from-port="18" to-layer="29" to-port="1" />
691
+ <edge from-layer="28" from-port="19" to-layer="37" to-port="1" />
692
+ <edge from-layer="28" from-port="19" to-layer="33" to-port="0" />
693
+ <edge from-layer="28" from-port="19" to-layer="32" to-port="0" />
694
+ <edge from-layer="28" from-port="19" to-layer="29" to-port="0" />
695
+ <edge from-layer="29" from-port="2" to-layer="31" to-port="0" />
696
+ <edge from-layer="30" from-port="0" to-layer="31" to-port="1" />
697
+ <edge from-layer="31" from-port="2" to-layer="32" to-port="1" />
698
+ <edge from-layer="32" from-port="2" to-layer="33" to-port="1" />
699
+ <edge from-layer="32" from-port="2" to-layer="37" to-port="0" />
700
+ <edge from-layer="33" from-port="2" to-layer="35" to-port="0" />
701
+ <edge from-layer="34" from-port="0" to-layer="35" to-port="1" />
702
+ <edge from-layer="35" from-port="2" to-layer="37" to-port="3" />
703
+ <edge from-layer="36" from-port="0" to-layer="37" to-port="4" />
704
+ <edge from-layer="37" from-port="6" to-layer="38" to-port="0" />
705
+ <edge from-layer="37" from-port="5" to-layer="41" to-port="0" />
706
+ <edge from-layer="38" from-port="1" to-layer="39" to-port="0" />
707
+ <edge from-layer="39" from-port="1" to-layer="40" to-port="0" />
708
+ <edge from-layer="41" from-port="1" to-layer="42" to-port="0" />
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
709
  </edges>
710
  <rt_info>
711
  <eos_token_id value="50256" />
712
+ <original_tokenizer_class value="&lt;class 'transformers_modules.pytorch.tokenization_codegen25.CodeGen25Tokenizer'>" />
713
  </rt_info>
714
  </net>
tokenization_codegen25.py CHANGED
@@ -245,4 +245,4 @@ class CodeGen25Tokenizer(PreTrainedTokenizer):
245
 
246
  # has no vocab file
247
  def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None):
248
- return ()
 
245
 
246
  # has no vocab file
247
  def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None):
248
+ return ()
tokenizer_config.json CHANGED
@@ -18,7 +18,7 @@
18
  },
19
  "clean_up_tokenization_spaces": true,
20
  "eos_token": "<|endoftext|>",
21
- "model_max_length": 2048,
22
  "pad_token": null,
23
  "tokenizer_class": "CodeGen25Tokenizer"
24
  }
 
18
  },
19
  "clean_up_tokenization_spaces": true,
20
  "eos_token": "<|endoftext|>",
21
+ "model_max_length": 1000000000000000019884624838656,
22
  "pad_token": null,
23
  "tokenizer_class": "CodeGen25Tokenizer"
24
  }