cromz22 commited on
Commit
3665b39
1 Parent(s): aed331f

upload model files

Browse files
config.json ADDED
@@ -0,0 +1,293 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "SpeechEncoderDecoderModel"
4
+ ],
5
+ "decoder": {
6
+ "_name_or_path": "facebook/bart-base",
7
+ "activation_dropout": 0.1,
8
+ "activation_function": "gelu",
9
+ "add_bias_logits": false,
10
+ "add_cross_attention": true,
11
+ "add_final_layer_norm": false,
12
+ "architectures": [
13
+ "BartModel"
14
+ ],
15
+ "attention_dropout": 0.1,
16
+ "bad_words_ids": null,
17
+ "bos_token_id": 0,
18
+ "chunk_size_feed_forward": 0,
19
+ "classif_dropout": 0.1,
20
+ "classifier_dropout": 0.0,
21
+ "cross_attention_hidden_size": null,
22
+ "d_model": 768,
23
+ "decoder_attention_heads": 12,
24
+ "decoder_ffn_dim": 3072,
25
+ "decoder_layerdrop": 0.0,
26
+ "decoder_layers": 6,
27
+ "decoder_start_token_id": 2,
28
+ "diversity_penalty": 0.0,
29
+ "do_sample": false,
30
+ "dropout": 0.1,
31
+ "early_stopping": true,
32
+ "encoder_attention_heads": 12,
33
+ "encoder_ffn_dim": 3072,
34
+ "encoder_layerdrop": 0.0,
35
+ "encoder_layers": 6,
36
+ "encoder_no_repeat_ngram_size": 0,
37
+ "eos_token_id": 2,
38
+ "exponential_decay_length_penalty": null,
39
+ "finetuning_task": null,
40
+ "forced_bos_token_id": 0,
41
+ "forced_eos_token_id": 2,
42
+ "gradient_checkpointing": false,
43
+ "id2label": {
44
+ "0": "LABEL_0",
45
+ "1": "LABEL_1",
46
+ "2": "LABEL_2"
47
+ },
48
+ "init_std": 0.02,
49
+ "is_decoder": true,
50
+ "is_encoder_decoder": false,
51
+ "label2id": {
52
+ "LABEL_0": 0,
53
+ "LABEL_1": 1,
54
+ "LABEL_2": 2
55
+ },
56
+ "length_penalty": 1.0,
57
+ "max_length": 20,
58
+ "max_position_embeddings": 1024,
59
+ "min_length": 0,
60
+ "model_type": "bart",
61
+ "no_repeat_ngram_size": 3,
62
+ "normalize_before": false,
63
+ "normalize_embedding": true,
64
+ "num_beam_groups": 1,
65
+ "num_beams": 4,
66
+ "num_hidden_layers": 6,
67
+ "num_return_sequences": 1,
68
+ "output_attentions": false,
69
+ "output_hidden_states": false,
70
+ "output_scores": false,
71
+ "pad_token_id": 1,
72
+ "prefix": null,
73
+ "problem_type": null,
74
+ "pruned_heads": {},
75
+ "remove_invalid_values": false,
76
+ "repetition_penalty": 1.0,
77
+ "return_dict": true,
78
+ "return_dict_in_generate": false,
79
+ "scale_embedding": false,
80
+ "sep_token_id": null,
81
+ "task_specific_params": {
82
+ "summarization": {
83
+ "length_penalty": 1.0,
84
+ "max_length": 128,
85
+ "min_length": 12,
86
+ "num_beams": 4
87
+ },
88
+ "summarization_cnn": {
89
+ "length_penalty": 2.0,
90
+ "max_length": 142,
91
+ "min_length": 56,
92
+ "num_beams": 4
93
+ },
94
+ "summarization_xsum": {
95
+ "length_penalty": 1.0,
96
+ "max_length": 62,
97
+ "min_length": 11,
98
+ "num_beams": 6
99
+ }
100
+ },
101
+ "temperature": 1.0,
102
+ "tie_encoder_decoder": false,
103
+ "tie_word_embeddings": true,
104
+ "tokenizer_class": null,
105
+ "top_k": 50,
106
+ "top_p": 1.0,
107
+ "torch_dtype": "float32",
108
+ "torchscript": false,
109
+ "transformers_version": "4.20.0.dev0",
110
+ "typical_p": 1.0,
111
+ "use_bfloat16": false,
112
+ "use_cache": true,
113
+ "vocab_size": 50265
114
+ },
115
+ "decoder_start_token_id": 0,
116
+ "encoder": {
117
+ "_name_or_path": "facebook/wav2vec2-base",
118
+ "activation_dropout": 0.0,
119
+ "adapter_kernel_size": 3,
120
+ "adapter_stride": 2,
121
+ "add_adapter": true,
122
+ "add_cross_attention": false,
123
+ "apply_spec_augment": true,
124
+ "architectures": [
125
+ "Wav2Vec2ForPreTraining"
126
+ ],
127
+ "attention_dropout": 0.1,
128
+ "bad_words_ids": null,
129
+ "bos_token_id": 1,
130
+ "chunk_size_feed_forward": 0,
131
+ "classifier_proj_size": 256,
132
+ "codevector_dim": 256,
133
+ "contrastive_logits_temperature": 0.1,
134
+ "conv_bias": false,
135
+ "conv_dim": [
136
+ 512,
137
+ 512,
138
+ 512,
139
+ 512,
140
+ 512,
141
+ 512,
142
+ 512
143
+ ],
144
+ "conv_kernel": [
145
+ 10,
146
+ 3,
147
+ 3,
148
+ 3,
149
+ 3,
150
+ 2,
151
+ 2
152
+ ],
153
+ "conv_stride": [
154
+ 5,
155
+ 2,
156
+ 2,
157
+ 2,
158
+ 2,
159
+ 2,
160
+ 2
161
+ ],
162
+ "cross_attention_hidden_size": null,
163
+ "ctc_loss_reduction": "sum",
164
+ "ctc_zero_infinity": false,
165
+ "decoder_start_token_id": null,
166
+ "diversity_loss_weight": 0.1,
167
+ "diversity_penalty": 0.0,
168
+ "do_sample": false,
169
+ "do_stable_layer_norm": false,
170
+ "early_stopping": false,
171
+ "encoder_no_repeat_ngram_size": 0,
172
+ "eos_token_id": 2,
173
+ "exponential_decay_length_penalty": null,
174
+ "feat_extract_activation": "gelu",
175
+ "feat_extract_norm": "group",
176
+ "feat_proj_dropout": 0.0,
177
+ "feat_quantizer_dropout": 0.0,
178
+ "final_dropout": 0.0,
179
+ "finetuning_task": null,
180
+ "forced_bos_token_id": null,
181
+ "forced_eos_token_id": null,
182
+ "freeze_feat_extract_train": true,
183
+ "hidden_act": "gelu",
184
+ "hidden_dropout": 0.1,
185
+ "hidden_size": 768,
186
+ "id2label": {
187
+ "0": "LABEL_0",
188
+ "1": "LABEL_1"
189
+ },
190
+ "initializer_range": 0.02,
191
+ "intermediate_size": 3072,
192
+ "is_decoder": false,
193
+ "is_encoder_decoder": false,
194
+ "label2id": {
195
+ "LABEL_0": 0,
196
+ "LABEL_1": 1
197
+ },
198
+ "layer_norm_eps": 1e-05,
199
+ "layerdrop": 0.0,
200
+ "length_penalty": 1.0,
201
+ "mask_channel_length": 10,
202
+ "mask_channel_min_space": 1,
203
+ "mask_channel_other": 0.0,
204
+ "mask_channel_prob": 0.0,
205
+ "mask_channel_selection": "static",
206
+ "mask_feature_length": 10,
207
+ "mask_feature_min_masks": 0,
208
+ "mask_feature_prob": 0.0,
209
+ "mask_time_length": 10,
210
+ "mask_time_min_masks": 2,
211
+ "mask_time_min_space": 1,
212
+ "mask_time_other": 0.0,
213
+ "mask_time_prob": 0.05,
214
+ "mask_time_selection": "static",
215
+ "max_length": 20,
216
+ "min_length": 0,
217
+ "model_type": "wav2vec2",
218
+ "no_mask_channel_overlap": false,
219
+ "no_mask_time_overlap": false,
220
+ "no_repeat_ngram_size": 0,
221
+ "num_adapter_layers": 3,
222
+ "num_attention_heads": 12,
223
+ "num_beam_groups": 1,
224
+ "num_beams": 1,
225
+ "num_codevector_groups": 2,
226
+ "num_codevectors_per_group": 320,
227
+ "num_conv_pos_embedding_groups": 16,
228
+ "num_conv_pos_embeddings": 128,
229
+ "num_feat_extract_layers": 7,
230
+ "num_hidden_layers": 12,
231
+ "num_negatives": 100,
232
+ "num_return_sequences": 1,
233
+ "output_attentions": false,
234
+ "output_hidden_size": 768,
235
+ "output_hidden_states": false,
236
+ "output_scores": false,
237
+ "pad_token_id": 0,
238
+ "prefix": null,
239
+ "problem_type": null,
240
+ "proj_codevector_dim": 256,
241
+ "pruned_heads": {},
242
+ "remove_invalid_values": false,
243
+ "repetition_penalty": 1.0,
244
+ "return_dict": true,
245
+ "return_dict_in_generate": false,
246
+ "sep_token_id": null,
247
+ "task_specific_params": null,
248
+ "tdnn_dilation": [
249
+ 1,
250
+ 2,
251
+ 3,
252
+ 1,
253
+ 1
254
+ ],
255
+ "tdnn_dim": [
256
+ 512,
257
+ 512,
258
+ 512,
259
+ 512,
260
+ 1500
261
+ ],
262
+ "tdnn_kernel": [
263
+ 5,
264
+ 3,
265
+ 3,
266
+ 1,
267
+ 1
268
+ ],
269
+ "temperature": 1.0,
270
+ "tie_encoder_decoder": false,
271
+ "tie_word_embeddings": true,
272
+ "tokenizer_class": null,
273
+ "top_k": 50,
274
+ "top_p": 1.0,
275
+ "torch_dtype": null,
276
+ "torchscript": false,
277
+ "transformers_version": "4.20.0.dev0",
278
+ "typical_p": 1.0,
279
+ "use_bfloat16": false,
280
+ "use_weighted_layer_sum": false,
281
+ "vocab_size": 32,
282
+ "xvector_output_dim": 512
283
+ },
284
+ "eos_token_id": 2,
285
+ "is_encoder_decoder": true,
286
+ "max_length": 200,
287
+ "model_type": "speech-encoder-decoder",
288
+ "num_beams": 5,
289
+ "pad_token_id": 1,
290
+ "tie_word_embeddings": false,
291
+ "torch_dtype": "float32",
292
+ "transformers_version": null
293
+ }
load_and_save.py ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python3
2
+
3
+ # from transformers.utils.dummy_pt_objects import SpeechEncoderDecoderModel
4
+ # from transformers.models.auto.feature_extraction_auto import AutoFeatureExtractor
5
+ # from transformers.models.auto.tokenization_auto import AutoTokenizer
6
+ # from transformers.models.wav2vec2.processing_wav2vec2 import Wav2Vec2Processor
7
+
8
+ from transformers import SpeechEncoderDecoderModel, AutoFeatureExtractor, AutoTokenizer, Wav2Vec2Processor
9
+
10
+
11
+ # checkpoints to leverage
12
+ encoder_id = "facebook/wav2vec2-base"
13
+ decoder_id = "facebook/bart-base"
14
+
15
+ # load and save speech-encoder-decoder model
16
+ # set some hyper-parameters for training and evaluation
17
+ model = SpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
18
+ encoder_id,
19
+ decoder_id,
20
+ encoder_add_adapter=True,
21
+ encoder_feat_proj_dropout=0.0,
22
+ encoder_layerdrop=0.0,
23
+ max_length=200,
24
+ num_beams=5,
25
+ )
26
+ model.config.decoder_start_token_id = model.decoder.config.bos_token_id
27
+ model.config.pad_token_id = model.decoder.config.pad_token_id
28
+ model.config.eos_token_id = model.decoder.config.eos_token_id
29
+ model.save_pretrained("./")
30
+
31
+ # load and save processor
32
+ feature_extractor = AutoFeatureExtractor.from_pretrained(encoder_id)
33
+ tokenizer = AutoTokenizer.from_pretrained(decoder_id)
34
+ processor = Wav2Vec2Processor(feature_extractor, tokenizer)
35
+ processor.save_pretrained("./")
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
preprocessor_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0.0,
7
+ "processor_class": "Wav2Vec2Processor",
8
+ "return_attention_mask": false,
9
+ "sampling_rate": 16000
10
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8c5e3f1f9ce8bdd2393f46a380784072a9fd50a1d61f21cf5876956a712a79d
3
+ size 804509967
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"errors": "replace", "bos_token": "<s>", "eos_token": "</s>", "sep_token": "</s>", "cls_token": "<s>", "unk_token": "<unk>", "pad_token": "<pad>", "mask_token": "<mask>", "add_prefix_space": false, "trim_offsets": true, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "facebook/bart-base", "tokenizer_class": "BartTokenizer", "processor_class": "Wav2Vec2Processor"}
vocab.json ADDED
The diff for this file is too large to render. See raw diff