sanchit-gandhi HF staff commited on
Commit
8f8f0d5
1 Parent(s): 561e2ac
.gitattributes CHANGED
@@ -25,3 +25,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
25
  *.zip filter=lfs diff=lfs merge=lfs -text
26
  *.zstandard filter=lfs diff=lfs merge=lfs -text
27
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
25
  *.zip filter=lfs diff=lfs merge=lfs -text
26
  *.zstandard filter=lfs diff=lfs merge=lfs -text
27
  *tfevents* filter=lfs diff=lfs merge=lfs -text
28
+ *.wandb filter=lfs diff=lfs merge=lfs -text
29
+ nohup.out filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/README.md
config.json ADDED
@@ -0,0 +1,289 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "sanchit-gandhi/flax-wav2vec2-2-bart-large-scan",
3
+ "architectures": [
4
+ "SpeechEncoderDecoderModel"
5
+ ],
6
+ "decoder": {
7
+ "_name_or_path": "",
8
+ "activation_dropout": 0.1,
9
+ "activation_function": "gelu",
10
+ "add_bias_logits": false,
11
+ "add_cross_attention": true,
12
+ "add_final_layer_norm": false,
13
+ "architectures": [
14
+ "BartModel"
15
+ ],
16
+ "attention_dropout": 0.1,
17
+ "bad_words_ids": null,
18
+ "bos_token_id": 0,
19
+ "chunk_size_feed_forward": 0,
20
+ "classif_dropout": 0.1,
21
+ "classifier_dropout": 0.0,
22
+ "cross_attention_hidden_size": null,
23
+ "d_model": 1024,
24
+ "decoder_attention_heads": 16,
25
+ "decoder_ffn_dim": 4096,
26
+ "decoder_layerdrop": 0.0,
27
+ "decoder_layers": 12,
28
+ "decoder_start_token_id": 2,
29
+ "diversity_penalty": 0.0,
30
+ "do_sample": false,
31
+ "dropout": 0.1,
32
+ "early_stopping": true,
33
+ "encoder_attention_heads": 16,
34
+ "encoder_ffn_dim": 4096,
35
+ "encoder_layerdrop": 0.0,
36
+ "encoder_layers": 12,
37
+ "encoder_no_repeat_ngram_size": 0,
38
+ "eos_token_id": 2,
39
+ "exponential_decay_length_penalty": null,
40
+ "finetuning_task": null,
41
+ "forced_bos_token_id": 0,
42
+ "forced_eos_token_id": 2,
43
+ "fuse_matmuls": false,
44
+ "gradient_checkpointing": true,
45
+ "id2label": {
46
+ "0": "LABEL_0",
47
+ "1": "LABEL_1",
48
+ "2": "LABEL_2"
49
+ },
50
+ "init_std": 0.02,
51
+ "is_decoder": true,
52
+ "is_encoder_decoder": false,
53
+ "label2id": {
54
+ "LABEL_0": 0,
55
+ "LABEL_1": 1,
56
+ "LABEL_2": 2
57
+ },
58
+ "length_penalty": 1.0,
59
+ "max_length": 20,
60
+ "max_position_embeddings": 1024,
61
+ "min_length": 0,
62
+ "model_type": "bart",
63
+ "no_repeat_ngram_size": 3,
64
+ "normalize_before": false,
65
+ "num_beam_groups": 1,
66
+ "num_beams": 4,
67
+ "num_hidden_layers": 12,
68
+ "num_return_sequences": 1,
69
+ "output_attentions": false,
70
+ "output_hidden_states": false,
71
+ "output_scores": false,
72
+ "pad_token_id": 1,
73
+ "prefix": null,
74
+ "problem_type": null,
75
+ "pruned_heads": {},
76
+ "remove_invalid_values": false,
77
+ "repetition_penalty": 1.0,
78
+ "return_dict": true,
79
+ "return_dict_in_generate": false,
80
+ "scale_embedding": false,
81
+ "sep_token_id": null,
82
+ "task_specific_params": {
83
+ "summarization": {
84
+ "length_penalty": 1.0,
85
+ "max_length": 128,
86
+ "min_length": 12,
87
+ "num_beams": 4
88
+ },
89
+ "summarization_cnn": {
90
+ "length_penalty": 2.0,
91
+ "max_length": 142,
92
+ "min_length": 56,
93
+ "num_beams": 4
94
+ },
95
+ "summarization_xsum": {
96
+ "length_penalty": 1.0,
97
+ "max_length": 62,
98
+ "min_length": 11,
99
+ "num_beams": 6
100
+ }
101
+ },
102
+ "temperature": 1.0,
103
+ "tie_encoder_decoder": false,
104
+ "tie_word_embeddings": true,
105
+ "tokenizer_class": null,
106
+ "top_k": 50,
107
+ "top_p": 1.0,
108
+ "torch_dtype": "float32",
109
+ "torchscript": false,
110
+ "transformers_version": "4.18.0.dev0",
111
+ "typical_p": 1.0,
112
+ "use_bfloat16": false,
113
+ "use_cache": true,
114
+ "use_scan": true,
115
+ "vocab_size": 50265
116
+ },
117
+ "decoder_start_token_id": 0,
118
+ "encoder": {
119
+ "_name_or_path": "",
120
+ "activation_dropout": 0.1,
121
+ "adapter_kernel_size": 3,
122
+ "adapter_stride": 2,
123
+ "add_adapter": true,
124
+ "add_cross_attention": false,
125
+ "apply_spec_augment": true,
126
+ "architectures": [
127
+ "Wav2Vec2ForPreTraining"
128
+ ],
129
+ "attention_dropout": 0.1,
130
+ "bad_words_ids": null,
131
+ "bos_token_id": 1,
132
+ "chunk_size_feed_forward": 0,
133
+ "classifier_proj_size": 256,
134
+ "codevector_dim": 768,
135
+ "contrastive_logits_temperature": 0.1,
136
+ "conv_bias": true,
137
+ "conv_dim": [
138
+ 512,
139
+ 512,
140
+ 512,
141
+ 512,
142
+ 512,
143
+ 512,
144
+ 512
145
+ ],
146
+ "conv_kernel": [
147
+ 10,
148
+ 3,
149
+ 3,
150
+ 3,
151
+ 3,
152
+ 2,
153
+ 2
154
+ ],
155
+ "conv_stride": [
156
+ 5,
157
+ 2,
158
+ 2,
159
+ 2,
160
+ 2,
161
+ 2,
162
+ 2
163
+ ],
164
+ "cross_attention_hidden_size": null,
165
+ "ctc_loss_reduction": "sum",
166
+ "ctc_zero_infinity": false,
167
+ "decoder_start_token_id": null,
168
+ "diversity_loss_weight": 0.1,
169
+ "diversity_penalty": 0.0,
170
+ "do_sample": false,
171
+ "do_stable_layer_norm": true,
172
+ "early_stopping": false,
173
+ "encoder_no_repeat_ngram_size": 0,
174
+ "eos_token_id": 2,
175
+ "exponential_decay_length_penalty": null,
176
+ "feat_extract_activation": "gelu",
177
+ "feat_extract_dropout": 0.0,
178
+ "feat_extract_norm": "layer",
179
+ "feat_proj_dropout": 0.0,
180
+ "feat_quantizer_dropout": 0.0,
181
+ "final_dropout": 0.0,
182
+ "finetuning_task": null,
183
+ "forced_bos_token_id": null,
184
+ "forced_eos_token_id": null,
185
+ "fuse_matmuls": false,
186
+ "gradient_checkpointing": true,
187
+ "hidden_act": "gelu",
188
+ "hidden_dropout": 0.1,
189
+ "hidden_dropout_prob": 0.1,
190
+ "hidden_size": 1024,
191
+ "id2label": {
192
+ "0": "LABEL_0",
193
+ "1": "LABEL_1"
194
+ },
195
+ "initializer_range": 0.02,
196
+ "intermediate_size": 4096,
197
+ "is_decoder": false,
198
+ "is_encoder_decoder": false,
199
+ "label2id": {
200
+ "LABEL_0": 0,
201
+ "LABEL_1": 1
202
+ },
203
+ "layer_norm_eps": 1e-05,
204
+ "layerdrop": 0.0,
205
+ "length_penalty": 1.0,
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_prob": 0.1,
212
+ "max_length": 20,
213
+ "min_length": 0,
214
+ "model_type": "wav2vec2",
215
+ "no_repeat_ngram_size": 0,
216
+ "num_adapter_layers": 3,
217
+ "num_attention_heads": 16,
218
+ "num_beam_groups": 1,
219
+ "num_beams": 1,
220
+ "num_codevector_groups": 2,
221
+ "num_codevectors_per_group": 320,
222
+ "num_conv_pos_embedding_groups": 16,
223
+ "num_conv_pos_embeddings": 128,
224
+ "num_feat_extract_layers": 7,
225
+ "num_hidden_layers": 24,
226
+ "num_negatives": 100,
227
+ "num_return_sequences": 1,
228
+ "output_attentions": false,
229
+ "output_hidden_size": 1024,
230
+ "output_hidden_states": false,
231
+ "output_scores": false,
232
+ "pad_token_id": 0,
233
+ "prefix": null,
234
+ "problem_type": null,
235
+ "proj_codevector_dim": 768,
236
+ "pruned_heads": {},
237
+ "remove_invalid_values": false,
238
+ "repetition_penalty": 1.0,
239
+ "return_dict": true,
240
+ "return_dict_in_generate": false,
241
+ "sep_token_id": null,
242
+ "task_specific_params": null,
243
+ "tdnn_dilation": [
244
+ 1,
245
+ 2,
246
+ 3,
247
+ 1,
248
+ 1
249
+ ],
250
+ "tdnn_dim": [
251
+ 512,
252
+ 512,
253
+ 512,
254
+ 512,
255
+ 1500
256
+ ],
257
+ "tdnn_kernel": [
258
+ 5,
259
+ 3,
260
+ 3,
261
+ 1,
262
+ 1
263
+ ],
264
+ "temperature": 1.0,
265
+ "tie_encoder_decoder": false,
266
+ "tie_word_embeddings": true,
267
+ "tokenizer_class": null,
268
+ "top_k": 50,
269
+ "top_p": 1.0,
270
+ "torch_dtype": null,
271
+ "torchscript": false,
272
+ "transformers_version": "4.18.0.dev0",
273
+ "typical_p": 1.0,
274
+ "use_bfloat16": false,
275
+ "use_scan": true,
276
+ "use_weighted_layer_sum": false,
277
+ "vocab_size": 32,
278
+ "xvector_output_dim": 512
279
+ },
280
+ "eos_token_id": 2,
281
+ "is_encoder_decoder": true,
282
+ "max_length": 40,
283
+ "model_type": "speech-encoder-decoder",
284
+ "pad_token_id": 1,
285
+ "processor_class": "Wav2Vec2Processor",
286
+ "tie_word_embeddings": false,
287
+ "transformers_version": null,
288
+ "use_cache": false
289
+ }
get_ctc_tokenizer.py ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/get_ctc_tokenizer.py
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
models ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/models
nohup.out ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ff7c786b45a6845a8e77375046d4927ea7c18cea820a6b9c2f86061d6716f2f2
3
+ size 10931583
preprocessor_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0.0,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
run_ctc_dummy.sh ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/run_ctc_dummy.sh
run_flax_speech_recognition_ami_ctc.py ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/run_flax_speech_recognition_ami_ctc.py
run_flax_speech_recognition_ami_seq2seq.py ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/run_flax_speech_recognition_ami_seq2seq.py
run_flax_speech_recognition_ctc.py ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/run_flax_speech_recognition_ctc.py
run_flax_speech_recognition_ctc_ngram.py ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/run_flax_speech_recognition_ctc_ngram.py
run_flax_speech_recognition_seq2seq.py ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/run_flax_speech_recognition_seq2seq.py
run_seq2seq_dummy.sh ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/run_seq2seq_dummy.sh
seq2seq.png ADDED
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}}
tests ADDED
@@ -0,0 +1 @@
 
 
1
+ /home/sanchitgandhi/seq2seq-speech/tests
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": "sanchit-gandhi/flax-wav2vec2-2-bart-large-scan", "tokenizer_class": "BartTokenizer"}
vocab.json ADDED
The diff for this file is too large to render. See raw diff