Rolv-Arild commited on
Commit
5a2b856
·
1 Parent(s): 05ba5dd

Training in progress, step 500

Browse files
Files changed (40) hide show
  1. .gitattributes +1 -0
  2. .gitignore +1 -0
  3. added_tokens.json +1 -0
  4. config.json +107 -0
  5. eval.py +175 -0
  6. preprocessor_config.json +9 -0
  7. pytorch_model.bin +3 -0
  8. run.sh +39 -0
  9. run_speech_recognition_ctc.py +819 -0
  10. special_tokens_map.json +1 -0
  11. tokenizer_config.json +1 -0
  12. training_args.bin +3 -0
  13. vocab.json +1 -0
  14. wandb/debug-internal.log +1 -0
  15. wandb/debug.log +1 -0
  16. wandb/latest-run +1 -0
  17. wandb/run-20220523_091609-1iboydmy/files/config.yaml +0 -0
  18. wandb/run-20220523_091609-1iboydmy/files/output.log +1379 -0
  19. wandb/run-20220523_091609-1iboydmy/files/requirements.txt +77 -0
  20. wandb/run-20220523_091609-1iboydmy/files/wandb-metadata.json +62 -0
  21. wandb/run-20220523_091609-1iboydmy/files/wandb-summary.json +0 -0
  22. wandb/run-20220523_091609-1iboydmy/logs/debug-internal.log +0 -0
  23. wandb/run-20220523_091609-1iboydmy/logs/debug.log +189 -0
  24. wandb/run-20220523_091609-1iboydmy/run-1iboydmy.wandb +3 -0
  25. wandb/run-20220523_103002-wygrs7tw/files/config.yaml +0 -0
  26. wandb/run-20220523_103002-wygrs7tw/files/output.log +1746 -0
  27. wandb/run-20220523_103002-wygrs7tw/files/requirements.txt +77 -0
  28. wandb/run-20220523_103002-wygrs7tw/files/wandb-metadata.json +62 -0
  29. wandb/run-20220523_103002-wygrs7tw/files/wandb-summary.json +0 -0
  30. wandb/run-20220523_103002-wygrs7tw/logs/debug-internal.log +0 -0
  31. wandb/run-20220523_103002-wygrs7tw/logs/debug.log +181 -0
  32. wandb/run-20220523_103002-wygrs7tw/run-wygrs7tw.wandb +3 -0
  33. wandb/run-20220523_115145-3dybzmyz/files/config.yaml +0 -0
  34. wandb/run-20220523_115145-3dybzmyz/files/output.log +1788 -0
  35. wandb/run-20220523_115145-3dybzmyz/files/requirements.txt +77 -0
  36. wandb/run-20220523_115145-3dybzmyz/files/wandb-metadata.json +62 -0
  37. wandb/run-20220523_115145-3dybzmyz/files/wandb-summary.json +0 -0
  38. wandb/run-20220523_115145-3dybzmyz/logs/debug-internal.log +0 -0
  39. wandb/run-20220523_115145-3dybzmyz/logs/debug.log +27 -0
  40. wandb/run-20220523_115145-3dybzmyz/run-3dybzmyz.wandb +3 -0
.gitattributes CHANGED
@@ -25,3 +25,4 @@ 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
.gitignore ADDED
@@ -0,0 +1 @@
 
 
1
+ checkpoint-*/
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"<s>": 32, "</s>": 33}
config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/wav2vec2-xls-r-1b",
3
+ "activation_dropout": 0.055,
4
+ "adapter_kernel_size": 3,
5
+ "adapter_stride": 2,
6
+ "add_adapter": false,
7
+ "apply_spec_augment": true,
8
+ "architectures": [
9
+ "Wav2Vec2ForCTC"
10
+ ],
11
+ "attention_dropout": 0.094,
12
+ "bos_token_id": 1,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 1024,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": true,
17
+ "conv_dim": [
18
+ 512,
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512,
24
+ 512
25
+ ],
26
+ "conv_kernel": [
27
+ 10,
28
+ 3,
29
+ 3,
30
+ 3,
31
+ 3,
32
+ 2,
33
+ 2
34
+ ],
35
+ "conv_stride": [
36
+ 5,
37
+ 2,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2,
42
+ 2
43
+ ],
44
+ "ctc_loss_reduction": "mean",
45
+ "ctc_zero_infinity": true,
46
+ "diversity_loss_weight": 0.1,
47
+ "do_stable_layer_norm": true,
48
+ "eos_token_id": 2,
49
+ "feat_extract_activation": "gelu",
50
+ "feat_extract_dropout": 0.0,
51
+ "feat_extract_norm": "layer",
52
+ "feat_proj_dropout": 0.04,
53
+ "feat_quantizer_dropout": 0.0,
54
+ "final_dropout": 0.0,
55
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.047,
57
+ "hidden_size": 1280,
58
+ "initializer_range": 0.02,
59
+ "intermediate_size": 5120,
60
+ "layer_norm_eps": 1e-05,
61
+ "layerdrop": 0.041,
62
+ "mask_feature_length": 64,
63
+ "mask_feature_min_masks": 0,
64
+ "mask_feature_prob": 0.25,
65
+ "mask_time_length": 10,
66
+ "mask_time_min_masks": 2,
67
+ "mask_time_prob": 0.082,
68
+ "model_type": "wav2vec2",
69
+ "num_adapter_layers": 3,
70
+ "num_attention_heads": 16,
71
+ "num_codevector_groups": 2,
72
+ "num_codevectors_per_group": 320,
73
+ "num_conv_pos_embedding_groups": 16,
74
+ "num_conv_pos_embeddings": 128,
75
+ "num_feat_extract_layers": 7,
76
+ "num_hidden_layers": 48,
77
+ "num_negatives": 100,
78
+ "output_hidden_size": 1280,
79
+ "pad_token_id": 31,
80
+ "proj_codevector_dim": 1024,
81
+ "tdnn_dilation": [
82
+ 1,
83
+ 2,
84
+ 3,
85
+ 1,
86
+ 1
87
+ ],
88
+ "tdnn_dim": [
89
+ 512,
90
+ 512,
91
+ 512,
92
+ 512,
93
+ 1500
94
+ ],
95
+ "tdnn_kernel": [
96
+ 5,
97
+ 3,
98
+ 3,
99
+ 1,
100
+ 1
101
+ ],
102
+ "torch_dtype": "float32",
103
+ "transformers_version": "4.18.0",
104
+ "use_weighted_layer_sum": false,
105
+ "vocab_size": 34,
106
+ "xvector_output_dim": 512
107
+ }
eval.py ADDED
@@ -0,0 +1,175 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python3
2
+ import argparse
3
+ import re
4
+ from typing import Dict
5
+
6
+ import torch
7
+ from datasets import Audio, Dataset, load_dataset, load_metric
8
+
9
+ from transformers import AutoFeatureExtractor, pipeline, Wav2Vec2Processor, Wav2Vec2ProcessorWithLM, Wav2Vec2FeatureExtractor
10
+ from pyctcdecode import BeamSearchDecoderCTC
11
+
12
+
13
+ def log_results(result: Dataset, args: Dict[str, str]):
14
+ """DO NOT CHANGE. This function computes and logs the result metrics."""
15
+
16
+ log_outputs = args.log_outputs
17
+ lm = "withLM" if args.use_lm else "noLM"
18
+ model_id = args.model_id.replace("/", "_")
19
+ dataset_id = "_".join(args.dataset.split("/") + [model_id, args.config, args.split, lm])
20
+
21
+ # load metric
22
+ wer = load_metric("wer")
23
+ cer = load_metric("cer")
24
+
25
+ # compute metrics
26
+ wer_result = wer.compute(references=result["target"], predictions=result["prediction"])
27
+ cer_result = cer.compute(references=result["target"], predictions=result["prediction"])
28
+
29
+ # print & log results
30
+ result_str = f"WER: {wer_result}\n" f"CER: {cer_result}"
31
+ print(result_str)
32
+
33
+ with open(f"{dataset_id}_eval_results.txt", "w") as f:
34
+ f.write(result_str)
35
+
36
+ # log all results in text file. Possibly interesting for analysis
37
+ if log_outputs is not None:
38
+ pred_file = f"log_{dataset_id}_predictions.txt"
39
+ target_file = f"log_{dataset_id}_targets.txt"
40
+
41
+ with open(pred_file, "w") as p, open(target_file, "w") as t:
42
+ # mapping function to write output
43
+ def write_to_file(batch, i):
44
+ p.write(f"{i}" + "\n")
45
+ p.write(batch["prediction"] + "\n")
46
+ t.write(f"{i}" + "\n")
47
+ t.write(batch["target"] + "\n")
48
+
49
+ result.map(write_to_file, with_indices=True)
50
+
51
+
52
+ def normalize_text(text: str, dataset: str) -> str:
53
+ """DO ADAPT FOR YOUR USE CASE. this function normalizes the target text."""
54
+
55
+ chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\‘\”\�\'\–\_\\\+\#\/]' # noqa: W605 IMPORTANT: this should correspond to the chars that were ignored during training
56
+ text = re.sub(chars_to_ignore_regex, "", text.lower()) + " "
57
+
58
+ if dataset.lower().endswith("nst"):
59
+ text = text.lower()
60
+ text = text.replace("(...Vær stille under dette opptaket...)", "")
61
+ text = re.sub('[áàâ]', 'a', text)
62
+ text = re.sub('[ä]', 'æ', text)
63
+ text = re.sub('[éèëê]', 'e', text)
64
+ text = re.sub('[íìïî]', 'i', text)
65
+ text = re.sub('[óòöô]', 'o', text)
66
+ text = re.sub('[ö]', 'ø', text)
67
+ text = re.sub('[ç]', 'c', text)
68
+ text = re.sub('[úùüû]', 'u', text)
69
+ # text = re.sub('\\(?=(Punktum|Komma|Utropstegn|Spørsmålstegn))', ' ', text)
70
+ text = re.sub('\s+', ' ', text)
71
+ elif dataset.lower().endswith("npsc"):
72
+ text = re.sub('[áàâ]', 'a', text)
73
+ text = re.sub('[ä]', 'æ', text)
74
+ text = re.sub('[éèëê]', 'e', text)
75
+ text = re.sub('[íìïî]', 'i', text)
76
+ text = re.sub('[óòöô]', 'o', text)
77
+ text = re.sub('[ö]', 'ø', text)
78
+ text = re.sub('[ç]', 'c', text)
79
+ text = re.sub('[úùüû]', 'u', text)
80
+ text = re.sub('\s', ' ', text)
81
+ text = re.sub('<ee>', 'eee', text)
82
+ text = re.sub('<qq>', 'qqq', text)
83
+ text = re.sub('<mm>', 'mmm', text)
84
+ text = re.sub('<inaudible>', 'xxx', text)
85
+
86
+ # # In addition, we can normalize the target text, e.g. removing new lines characters etc...
87
+ # # note that order is important here!
88
+ # token_sequences_to_ignore = ["\n\n", "\n", " ", " "]
89
+
90
+ # for t in token_sequences_to_ignore:
91
+ # text = " ".join(text.split(t))
92
+
93
+ return text
94
+
95
+
96
+ def main(args):
97
+ # load dataset
98
+ dataset = load_dataset(args.dataset, args.config, split=args.split, use_auth_token=True)
99
+
100
+ # for testing: only process the first two examples as a test
101
+ # dataset = dataset.select(range(10))
102
+
103
+ # load processor
104
+ feature_extractor = AutoFeatureExtractor.from_pretrained(args.model_id)
105
+ sampling_rate = feature_extractor.sampling_rate
106
+
107
+ # resample audio
108
+ dataset = dataset.cast_column("audio", Audio(sampling_rate=sampling_rate))
109
+
110
+ # load eval pipeline
111
+ if args.device is None:
112
+ args.device = 0 if torch.cuda.is_available() else -1
113
+ # asr = pipeline("automatic-speech-recognition", model=args.model_id, device=args.device)
114
+
115
+ feature_extractor_dict, _ = Wav2Vec2FeatureExtractor.get_feature_extractor_dict(args.model_id)
116
+ feature_extractor_dict["processor_class"] = "Wav2Vec2Processor" if not args.use_lm else "Wav2Vec2ProcessorWithLM"
117
+ feature_extractor = Wav2Vec2FeatureExtractor.from_dict(feature_extractor_dict)
118
+
119
+ asr = pipeline("automatic-speech-recognition", model=args.model_id, feature_extractor=feature_extractor, device=args.device, decoder=BeamSearchDecoderCTC.load_from_dir("./"))
120
+
121
+ # map function to decode audio
122
+ def map_to_pred(batch):
123
+ prediction = asr(
124
+ batch["audio"]["array"], chunk_length_s=args.chunk_length_s, stride_length_s=args.stride_length_s
125
+ )
126
+
127
+ batch["prediction"] = prediction["text"]
128
+ batch["target"] = normalize_text(batch["text"], args.dataset)
129
+ return batch
130
+
131
+ # run inference on all examples
132
+ result = dataset.map(map_to_pred, remove_columns=dataset.column_names)
133
+
134
+ # compute and log_results
135
+ # do not change function below
136
+ log_results(result, args)
137
+
138
+
139
+ if __name__ == "__main__":
140
+ parser = argparse.ArgumentParser()
141
+
142
+ parser.add_argument(
143
+ "--model_id", type=str, required=True, help="Model identifier. Should be loadable with 🤗 Transformers"
144
+ )
145
+ parser.add_argument(
146
+ "--dataset",
147
+ type=str,
148
+ required=True,
149
+ help="Dataset name to evaluate the `model_id`. Should be loadable with 🤗 Datasets",
150
+ )
151
+ parser.add_argument(
152
+ "--config", type=str, required=True, help="Config of the dataset. *E.g.* `'en'` for Common Voice"
153
+ )
154
+ parser.add_argument("--split", type=str, required=True, help="Split of the dataset. *E.g.* `'test'`")
155
+ parser.add_argument(
156
+ "--chunk_length_s", type=float, default=None, help="Chunk length in seconds. Defaults to 5 seconds."
157
+ )
158
+ parser.add_argument(
159
+ "--stride_length_s", type=float, default=None, help="Stride of the audio chunks. Defaults to 1 second."
160
+ )
161
+ parser.add_argument(
162
+ "--log_outputs", action="store_true", help="If defined, write outputs to log file for analysis."
163
+ )
164
+ parser.add_argument(
165
+ "--device",
166
+ type=int,
167
+ default=None,
168
+ help="The device to run the pipeline on. -1 for CPU (default), 0 for the first GPU and so on.",
169
+ )
170
+ parser.add_argument(
171
+ "--use_lm", action="store_true", help="If defined, use included language model as the decoder."
172
+ )
173
+ args = parser.parse_args()
174
+
175
+ main(args)
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,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb89a06605f56711cff18daf05feca605389bdcb2628e4715f228b7f66767dd5
3
+ size 3850439217
run.sh ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ WANDB_ENTITY=NbAiLab WANDB_PROJECT=wav2vec2 python run_speech_recognition_ctc.py \
2
+ --model_name_or_path="facebook/wav2vec2-xls-r-1b" \
3
+ --hub_model_id="NbAiLab/wav2vec2-1b-npsc-nst-bokmaal" \
4
+ --output_dir="./" \
5
+ --overwrite_output_dir \
6
+ --num_train_epochs="40" \
7
+ --per_device_train_batch_size="12" \
8
+ --per_device_eval_batch_size="12" \
9
+ --gradient_accumulation_steps="2" \
10
+ --learning_rate="2e-5" \
11
+ --warmup_steps="2000" \
12
+ --length_column_name="input_length" \
13
+ --evaluation_strategy="steps" \
14
+ --text_column_name="text" \
15
+ --save_steps="500" \
16
+ --eval_steps="500" \
17
+ --logging_steps="100" \
18
+ --layerdrop="0.041" \
19
+ --attention_dropout="0.094" \
20
+ --activation_dropout="0.055" \
21
+ --hidden_dropout="0.047" \
22
+ --save_total_limit="3" \
23
+ --freeze_feature_encoder \
24
+ --feat_proj_dropout="0.04" \
25
+ --mask_time_prob="0.082" \
26
+ --mask_time_length="10" \
27
+ --mask_feature_prob="0.25" \
28
+ --mask_feature_length="64" \
29
+ --gradient_checkpointing \
30
+ --min_duration_in_seconds="0.5" \
31
+ --max_duration_in_seconds="30.0" \
32
+ --use_auth_token \
33
+ --seed="42" \
34
+ --fp16 \
35
+ --group_by_length \
36
+ --do_train --do_eval \
37
+ --push_to_hub \
38
+ --preprocessing_num_workers="32" \
39
+ --ctc_zero_infinity
run_speech_recognition_ctc.py ADDED
@@ -0,0 +1,819 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+ # coding=utf-8
3
+ # Copyright 2021 The HuggingFace Inc. team. All rights reserved.
4
+ #
5
+ # Licensed under the Apache License, Version 2.0 (the "License");
6
+ # you may not use this file except in compliance with the License.
7
+ # You may obtain a copy of the License at
8
+ #
9
+ # http://www.apache.org/licenses/LICENSE-2.0
10
+ #
11
+ # Unless required by applicable law or agreed to in writing, software
12
+ # distributed under the License is distributed on an "AS IS" BASIS,
13
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14
+ # See the License for the specific language governing permissions and
15
+
16
+ """ Fine-tuning a 🤗 Transformers CTC model for automatic speech recognition"""
17
+
18
+ import functools
19
+ import json
20
+ import logging
21
+ import os
22
+ import re
23
+ import sys
24
+ import warnings
25
+ from dataclasses import dataclass, field
26
+ from typing import Dict, List, Optional, Union
27
+
28
+ import datasets
29
+ import numpy as np
30
+ import torch
31
+ from datasets import DatasetDict, load_dataset, load_metric
32
+
33
+ import transformers
34
+ from transformers import (
35
+ AutoConfig,
36
+ AutoFeatureExtractor,
37
+ AutoModelForCTC,
38
+ AutoProcessor,
39
+ AutoTokenizer,
40
+ HfArgumentParser,
41
+ Trainer,
42
+ TrainingArguments,
43
+ Wav2Vec2Processor,
44
+ set_seed,
45
+ )
46
+ from transformers.trainer_utils import get_last_checkpoint, is_main_process
47
+ from transformers.utils import check_min_version
48
+ from transformers.utils.versions import require_version
49
+
50
+ # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
51
+ check_min_version("4.16.0.dev0")
52
+
53
+ require_version("datasets>=1.13.3", "To fix: pip install -r examples/pytorch/text-classification/requirements.txt")
54
+
55
+ logger = logging.getLogger(__name__)
56
+
57
+
58
+ def list_field(default=None, metadata=None):
59
+ return field(default_factory=lambda: default, metadata=metadata)
60
+
61
+
62
+ @dataclass
63
+ class ModelArguments:
64
+ """
65
+ Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
66
+ """
67
+
68
+ model_name_or_path: str = field(
69
+ metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
70
+ )
71
+ tokenizer_name_or_path: Optional[str] = field(
72
+ default=None,
73
+ metadata={"help": "Path to pretrained tokenizer or tokenizer identifier from huggingface.co/models"},
74
+ )
75
+ cache_dir: Optional[str] = field(
76
+ default=None,
77
+ metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"},
78
+ )
79
+ freeze_feature_encoder: bool = field(
80
+ default=True, metadata={"help": "Whether to freeze the feature encoder layers of the model."}
81
+ )
82
+ attention_dropout: float = field(
83
+ default=0.0, metadata={"help": "The dropout ratio for the attention probabilities."}
84
+ )
85
+ activation_dropout: float = field(
86
+ default=0.0, metadata={"help": "The dropout ratio for activations inside the fully connected layer."}
87
+ )
88
+ feat_proj_dropout: float = field(default=0.0, metadata={"help": "The dropout ratio for the projected features."})
89
+ hidden_dropout: float = field(
90
+ default=0.0,
91
+ metadata={
92
+ "help": "The dropout probability for all fully connected layers in the embeddings, encoder, and pooler."
93
+ },
94
+ )
95
+ final_dropout: float = field(
96
+ default=0.0,
97
+ metadata={"help": "The dropout probability for the final projection layer."},
98
+ )
99
+ mask_time_prob: float = field(
100
+ default=0.05,
101
+ metadata={
102
+ "help": "Probability of each feature vector along the time axis to be chosen as the start of the vector"
103
+ "span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"
104
+ "vectors will be masked along the time axis."
105
+ },
106
+ )
107
+ mask_time_length: int = field(
108
+ default=10,
109
+ metadata={"help": "Length of vector span to mask along the time axis."},
110
+ )
111
+ mask_feature_prob: float = field(
112
+ default=0.0,
113
+ metadata={
114
+ "help": "Probability of each feature vector along the feature axis to be chosen as the start of the vector"
115
+ "span to be masked. Approximately ``mask_feature_prob * sequence_length // mask_feature_length`` feature bins will be masked along the time axis."
116
+ },
117
+ )
118
+ mask_feature_length: int = field(
119
+ default=10,
120
+ metadata={"help": "Length of vector span to mask along the feature axis."},
121
+ )
122
+ layerdrop: float = field(default=0.0, metadata={"help": "The LayerDrop probability."})
123
+ ctc_loss_reduction: Optional[str] = field(
124
+ default="mean", metadata={"help": "The way the ctc loss should be reduced. Should be one of 'mean' or 'sum'."}
125
+ )
126
+ ctc_zero_infinity: Optional[bool] = field(
127
+ default=False, metadata={"help": "If True, will try yo aboud the CTC loss goinf to infinity."}
128
+ )
129
+
130
+
131
+ @dataclass
132
+ class DataTrainingArguments:
133
+ """
134
+ Arguments pertaining to what data we are going to input our model for training and eval.
135
+
136
+ Using `HfArgumentParser` we can turn this class
137
+ into argparse arguments to be able to specify them on
138
+ the command line.
139
+ """
140
+
141
+ # dataset_name: str = field(
142
+ # metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
143
+ # )
144
+ # dataset_config_name: str = field(
145
+ # default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
146
+ # )
147
+ train_split_name: str = field(
148
+ default="train",
149
+ metadata={
150
+ "help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
151
+ },
152
+ )
153
+ eval_split_name: str = field(
154
+ default="test",
155
+ metadata={
156
+ "help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
157
+ },
158
+ )
159
+ audio_column_name: str = field(
160
+ default="audio",
161
+ metadata={"help": "The name of the dataset column containing the audio data. Defaults to 'audio'"},
162
+ )
163
+ text_column_name: str = field(
164
+ default="text",
165
+ metadata={"help": "The name of the dataset column containing the text data. Defaults to 'text'"},
166
+ )
167
+ overwrite_cache: bool = field(
168
+ default=False, metadata={"help": "Overwrite the cached preprocessed datasets or not."}
169
+ )
170
+ preprocessing_num_workers: Optional[int] = field(
171
+ default=None,
172
+ metadata={"help": "The number of processes to use for the preprocessing."},
173
+ )
174
+ max_train_samples: Optional[int] = field(
175
+ default=None,
176
+ metadata={
177
+ "help": "For debugging purposes or quicker training, truncate the number of training examples to this "
178
+ "value if set."
179
+ },
180
+ )
181
+ max_eval_samples: Optional[int] = field(
182
+ default=None,
183
+ metadata={
184
+ "help": "For debugging purposes or quicker training, truncate the number of validation examples to this "
185
+ "value if set."
186
+ },
187
+ )
188
+ chars_to_ignore: Optional[List[str]] = list_field(
189
+ default=None,
190
+ metadata={"help": "A list of characters to remove from the transcripts."},
191
+ )
192
+ eval_metrics: List[str] = list_field(
193
+ default=["wer"],
194
+ metadata={"help": "A list of metrics the model should be evaluated on. E.g. `'wer cer'`"},
195
+ )
196
+ max_duration_in_seconds: float = field(
197
+ default=20.0,
198
+ metadata={
199
+ "help": "Filter audio files that are longer than `max_duration_in_seconds` seconds to 'max_duration_in_seconds`"
200
+ },
201
+ )
202
+ min_duration_in_seconds: float = field(
203
+ default=0.0, metadata={"help": "Filter audio files that are shorter than `min_duration_in_seconds` seconds"}
204
+ )
205
+ preprocessing_only: bool = field(
206
+ default=False,
207
+ metadata={
208
+ "help": "Whether to only do data preprocessing and skip training. "
209
+ "This is especially useful when data preprocessing errors out in distributed training due to timeout. "
210
+ "In this case, one should run the preprocessing in a non-distributed setup with `preprocessing_only=True` "
211
+ "so that the cached datasets can consequently be loaded in distributed training"
212
+ },
213
+ )
214
+ use_auth_token: bool = field(
215
+ default=False,
216
+ metadata={
217
+ "help": "If :obj:`True`, will use the token generated when running"
218
+ ":obj:`transformers-cli login` as HTTP bearer authorization for remote files."
219
+ },
220
+ )
221
+ unk_token: str = field(
222
+ default="[UNK]",
223
+ metadata={"help": "The unk token for the tokenizer"},
224
+ )
225
+ pad_token: str = field(
226
+ default="[PAD]",
227
+ metadata={"help": "The padding token for the tokenizer"},
228
+ )
229
+ word_delimiter_token: str = field(
230
+ default="|",
231
+ metadata={"help": "The word delimiter token for the tokenizer"},
232
+ )
233
+ phoneme_language: Optional[str] = field(
234
+ default=None,
235
+ metadata={
236
+ "help": "The target language that should be used be"
237
+ " passed to the tokenizer for tokenization. Note that"
238
+ " this is only relevant if the model classifies the"
239
+ " input audio to a sequence of phoneme sequences."
240
+ },
241
+ )
242
+
243
+
244
+ @dataclass
245
+ class DataCollatorCTCWithPadding:
246
+ """
247
+ Data collator that will dynamically pad the inputs received.
248
+ Args:
249
+ processor (:class:`~transformers.AutoProcessor`)
250
+ The processor used for proccessing the data.
251
+ padding (:obj:`bool`, :obj:`str` or :class:`~transformers.tokenization_utils_base.PaddingStrategy`, `optional`, defaults to :obj:`True`):
252
+ Select a strategy to pad the returned sequences (according to the model's padding side and padding index)
253
+ among:
254
+ * :obj:`True` or :obj:`'longest'`: Pad to the longest sequence in the batch (or no padding if only a single
255
+ sequence if provided).
256
+ * :obj:`'max_length'`: Pad to a maximum length specified with the argument :obj:`max_length` or to the
257
+ maximum acceptable input length for the model if that argument is not provided.
258
+ * :obj:`False` or :obj:`'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
259
+ different lengths).
260
+ max_length (:obj:`int`, `optional`):
261
+ Maximum length of the ``input_values`` of the returned list and optionally padding length (see above).
262
+ max_length_labels (:obj:`int`, `optional`):
263
+ Maximum length of the ``labels`` returned list and optionally padding length (see above).
264
+ pad_to_multiple_of (:obj:`int`, `optional`):
265
+ If set will pad the sequence to a multiple of the provided value.
266
+ This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >=
267
+ 7.5 (Volta).
268
+ """
269
+
270
+ processor: AutoProcessor
271
+ padding: Union[bool, str] = "longest"
272
+ pad_to_multiple_of: Optional[int] = None
273
+ pad_to_multiple_of_labels: Optional[int] = None
274
+
275
+ def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
276
+ # split inputs and labels since they have to be of different lenghts and need
277
+ # different padding methods
278
+ input_features = [{"input_values": feature["input_values"]} for feature in features]
279
+ label_features = [{"input_ids": feature["labels"]} for feature in features]
280
+
281
+ batch = self.processor.pad(
282
+ input_features,
283
+ padding=self.padding,
284
+ pad_to_multiple_of=self.pad_to_multiple_of,
285
+ return_tensors="pt",
286
+ )
287
+
288
+ with self.processor.as_target_processor():
289
+ labels_batch = self.processor.pad(
290
+ label_features,
291
+ padding=self.padding,
292
+ pad_to_multiple_of=self.pad_to_multiple_of_labels,
293
+ return_tensors="pt",
294
+ )
295
+
296
+ # replace padding with -100 to ignore loss correctly
297
+ labels = labels_batch["input_ids"].masked_fill(labels_batch.attention_mask.ne(1), -100)
298
+
299
+ batch["labels"] = labels
300
+
301
+ return batch
302
+
303
+
304
+ def create_vocabulary_from_data(
305
+ datasets: DatasetDict,
306
+ word_delimiter_token: Optional[str] = None,
307
+ unk_token: Optional[str] = None,
308
+ pad_token: Optional[str] = None,
309
+ ):
310
+ # Given training and test labels create vocabulary
311
+ alphabet = set()
312
+
313
+ def extract_all_chars(batch):
314
+ all_text = " ".join(batch["target_text"])
315
+ alphabet.update(all_text)
316
+
317
+ datasets.map(
318
+ extract_all_chars,
319
+ batched=True,
320
+ batch_size=-1,
321
+ keep_in_memory=True,
322
+ remove_columns=datasets["train"].column_names,
323
+ )
324
+
325
+ # # take union of all unique characters in each dataset
326
+ # vocab_set = functools.reduce(
327
+ # lambda vocab_1, vocab_2: {"vocab": list(set(vocab_1["vocab"][0]) | set(vocab_2["vocab"][0]))}, vocabs.values()
328
+ # )["vocab"][0]
329
+
330
+ vocab_dict = {v: k for k, v in enumerate(sorted(list(alphabet)))}
331
+
332
+ # replace white space with delimiter token
333
+ if word_delimiter_token is not None:
334
+ vocab_dict[word_delimiter_token] = vocab_dict[" "]
335
+ del vocab_dict[" "]
336
+
337
+ # add unk and pad token
338
+ if unk_token is not None:
339
+ vocab_dict[unk_token] = len(vocab_dict)
340
+
341
+ if pad_token is not None:
342
+ vocab_dict[pad_token] = len(vocab_dict)
343
+
344
+ return vocab_dict
345
+
346
+
347
+ def make_dataset(seed=42):
348
+ # Pre-processing dataset
349
+ import re
350
+
351
+ def replace_strange_characters(text):
352
+ text = re.sub('[áàâ]', 'a', text)
353
+ text = re.sub('[ä]', 'æ', text)
354
+ text = re.sub('[éèëê]', 'e', text)
355
+ text = re.sub('[íìïî]', 'i', text)
356
+ text = re.sub('[óòöô]', 'o', text)
357
+ text = re.sub('[ö]', 'ø', text)
358
+ text = re.sub('[ç]', 'c', text)
359
+ text = re.sub('[úùüû]', 'u', text)
360
+ return text
361
+
362
+ def replace_hesitations(text):
363
+ # text = re.sub("<[^>]*>", " ", text) # <ee>, <qq>, <mm>, <inaudible> for NPSC. <eeeh>, <mmm> for NST-hesitate
364
+ text = re.sub("<ee(eh)?>", "E", text)
365
+ text = re.sub("<mmm?>", "M", text)
366
+ text = re.sub("<qq>", "Q", text)
367
+ text = re.sub("<inaudible>", "I", text)
368
+ return text
369
+
370
+ def is_too_short(entry):
371
+ return len(entry["text"]) > len(entry["audio"]["array"]) // 320 or len(entry["text"]) <=1
372
+
373
+ def map_nst(entry):
374
+ text = entry["text"].lower()
375
+ text = text.replace("(...vær stille under dette opptaket...)", " ")
376
+ text = replace_hesitations(text)
377
+ text = replace_strange_characters(text)
378
+ text = re.sub('\s+', ' ', text)
379
+ return {"text": text.strip()}
380
+
381
+ def filter_nst(entry):
382
+ if is_too_short(entry):
383
+ return False # Too short
384
+ if re.match(entry["type"], "pIW|CA"):
385
+ return False # Spelling out words
386
+ if re.search("\d", entry["text"]):
387
+ return False
388
+ return True
389
+
390
+ def filter_npsc(entry):
391
+ if is_too_short(entry):
392
+ return False # Too short
393
+ if re.search("\d", entry["text"]):
394
+ return False
395
+ return True
396
+
397
+ def map_npsc(entry):
398
+ text = entry["text"] if entry["sentence_language_code"].startswith("nn") else entry["normsentence_text"]
399
+ text = text.lower()
400
+ text = replace_strange_characters(text)
401
+ text = replace_hesitations(text)
402
+ text = re.sub('\s+', ' ', text)
403
+ return {"text": text.strip()}
404
+
405
+ nst = datasets.load_dataset("NbAiLab/NST", "no-close")
406
+ npsc = datasets.load_dataset("NbAiLab/NPSC", "16K_mp3")
407
+ nsth = datasets.load_dataset("NbAiLab/NST_hesitate", "no")
408
+
409
+ nst = nst.map(map_nst).filter(filter_nst)
410
+ npsc = npsc.map(map_npsc).filter(filter_npsc)
411
+ nsth = nsth.map(map_nst).filter(filter_npsc)
412
+
413
+ split = len(npsc["train"]) / (len(npsc["train"]) + len(npsc["validation"])) # Use same train/val ratio as NPSC
414
+ nst_train = nst["train"].train_test_split(train_size=split, seed=seed)
415
+ nst["train"] = nst_train["train"]
416
+ nst["validation"] = nst_train["test"]
417
+
418
+ nsth_train = nsth["train"].train_test_split(train_size=split, seed=seed)
419
+ nsth["train"] = nsth_train["train"]
420
+ nsth["validation"] = nsth_train["test"]
421
+
422
+ nst_base = nst.remove_columns([col for col in nst["train"].column_names if col not in ["text", "audio"]])
423
+ npsc_base = npsc.remove_columns([col for col in npsc["train"].column_names if col not in ["text", "audio"]])
424
+ nsth_base = nsth.remove_columns([col for col in nsth["train"].column_names if col not in ["text", "audio"]])
425
+
426
+ combined = {}
427
+ for split in "train", "validation", "test":
428
+ # Weight by number of examples
429
+ probs = np.array([len(nst_base[split]), len(npsc_base[split]), len(nsth_base[split])])
430
+ probs = (probs / probs.sum()).tolist()
431
+ comb = datasets.interleave_datasets([nst_base[split], npsc_base[split], nsth_base[split]],
432
+ probabilities=probs, seed=seed)
433
+ combined[split] = comb
434
+
435
+ return datasets.DatasetDict(**combined)
436
+
437
+
438
+ def main():
439
+ # See all possible arguments in src/transformers/training_args.py
440
+ # or by passing the --help flag to this script.
441
+ # We now keep distinct sets of args, for a cleaner separation of concerns.
442
+
443
+ parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
444
+ if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
445
+ # If we pass only one argument to the script and it's the path to a json file,
446
+ # let's parse it to get our arguments.
447
+ model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
448
+ else:
449
+ model_args, data_args, training_args = parser.parse_args_into_dataclasses()
450
+
451
+ # Detecting last checkpoint.
452
+ last_checkpoint = None
453
+ if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
454
+ last_checkpoint = get_last_checkpoint(training_args.output_dir)
455
+ if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
456
+ raise ValueError(
457
+ f"Output directory ({training_args.output_dir}) already exists and is not empty. "
458
+ "Use --overwrite_output_dir to overcome."
459
+ )
460
+ elif last_checkpoint is not None:
461
+ logger.info(
462
+ f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
463
+ "the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
464
+ )
465
+
466
+ # Setup logging
467
+ logging.basicConfig(
468
+ format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
469
+ datefmt="%m/%d/%Y %H:%M:%S",
470
+ handlers=[logging.StreamHandler(sys.stdout)],
471
+ )
472
+ logger.setLevel(logging.INFO if is_main_process(training_args.local_rank) else logging.WARN)
473
+
474
+ # Log on each process the small summary:
475
+ logger.warning(
476
+ f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
477
+ f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
478
+ )
479
+ # Set the verbosity to info of the Transformers logger (on main process only):
480
+ if is_main_process(training_args.local_rank):
481
+ transformers.utils.logging.set_verbosity_info()
482
+ logger.info("Training/evaluation parameters %s", training_args)
483
+
484
+ # Set seed before initializing model.
485
+ set_seed(training_args.seed)
486
+
487
+ # 1. First, let's load the dataset
488
+ raw_datasets = make_dataset(seed=training_args.seed)
489
+
490
+ if training_args.do_train:
491
+ if data_args.audio_column_name not in raw_datasets["train"].column_names:
492
+ raise ValueError(
493
+ f"--audio_column_name '{data_args.audio_column_name}' not found in dataset '{data_args.dataset_name}'. "
494
+ "Make sure to set `--audio_column_name` to the correct audio column - one of "
495
+ f"{', '.join(raw_datasets['train'].column_names)}."
496
+ )
497
+
498
+ if data_args.text_column_name not in raw_datasets["train"].column_names:
499
+ raise ValueError(
500
+ f"--text_column_name {data_args.text_column_name} not found in dataset '{data_args.dataset_name}'. "
501
+ "Make sure to set `--text_column_name` to the correct text column - one of "
502
+ f"{', '.join(raw_datasets['train'].column_names)}."
503
+ )
504
+
505
+ if data_args.max_train_samples is not None:
506
+ raw_datasets["train"] = raw_datasets["train"].select(range(data_args.max_train_samples))
507
+
508
+ if training_args.do_eval:
509
+ if data_args.max_eval_samples is not None:
510
+ raw_datasets["eval"] = raw_datasets["eval"].select(range(data_args.max_eval_samples))
511
+
512
+ # 2. We remove some special characters from the datasets
513
+ # that make training complicated and do not help in transcribing the speech
514
+ # E.g. characters, such as `,` and `.` do not really have an acoustic characteristic
515
+ # that could be easily picked up by the model
516
+ # chars_to_ignore_regex = (
517
+ # f'[{"".join(data_args.chars_to_ignore)}]' if data_args.chars_to_ignore is not None else None
518
+ # )
519
+ chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\‘\”\�\'\–\_\\\+\#\/]'
520
+
521
+ text_column_name = data_args.text_column_name
522
+
523
+ def remove_special_characters(batch):
524
+ if chars_to_ignore_regex is not None:
525
+ batch["target_text"] = re.sub(chars_to_ignore_regex, "", batch[text_column_name]).lower() + " "
526
+ else:
527
+ batch["target_text"] = batch[text_column_name].lower() + " "
528
+ return batch
529
+
530
+ with training_args.main_process_first(desc="dataset map special characters removal"):
531
+ raw_datasets = raw_datasets.map(
532
+ remove_special_characters,
533
+ remove_columns=[text_column_name],
534
+ desc="remove special characters from datasets",
535
+ )
536
+
537
+ # save special tokens for tokenizer
538
+ word_delimiter_token = data_args.word_delimiter_token
539
+ unk_token = data_args.unk_token
540
+ pad_token = data_args.pad_token
541
+
542
+ # 3. Next, let's load the config as we might need it to create
543
+ # the tokenizer
544
+ # load config
545
+ config = AutoConfig.from_pretrained(
546
+ model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_auth_token=data_args.use_auth_token
547
+ )
548
+
549
+ # 4. Next, if no tokenizer file is defined,
550
+ # we create the vocabulary of the model by extracting all unique characters from
551
+ # the training and evaluation datasets
552
+ # We need to make sure that only first rank saves vocabulary
553
+ # make sure all processes wait until vocab is created
554
+ tokenizer_name_or_path = model_args.tokenizer_name_or_path
555
+ tokenizer_kwargs = {}
556
+ if tokenizer_name_or_path is None:
557
+ # save vocab in training output dir
558
+ tokenizer_name_or_path = training_args.output_dir
559
+
560
+ vocab_file = os.path.join(tokenizer_name_or_path, "vocab.json")
561
+
562
+ with training_args.main_process_first():
563
+ if training_args.overwrite_output_dir and os.path.isfile(vocab_file):
564
+ os.remove(vocab_file)
565
+
566
+ with training_args.main_process_first(desc="dataset map vocabulary creation"):
567
+ if not os.path.isfile(vocab_file):
568
+ os.makedirs(tokenizer_name_or_path, exist_ok=True)
569
+ vocab_dict = create_vocabulary_from_data(
570
+ raw_datasets,
571
+ word_delimiter_token=word_delimiter_token,
572
+ unk_token=unk_token,
573
+ pad_token=pad_token,
574
+ )
575
+
576
+ # save vocab dict to be loaded into tokenizer
577
+ with open(vocab_file, "w") as file:
578
+ json.dump(vocab_dict, file)
579
+
580
+ # if tokenizer has just been created
581
+ # it is defined by `tokenizer_class` if present in config else by `model_type`
582
+ tokenizer_kwargs = {
583
+ "config": config if config.tokenizer_class is not None else None,
584
+ "tokenizer_type": config.model_type if config.tokenizer_class is None else None,
585
+ "unk_token": unk_token,
586
+ "pad_token": pad_token,
587
+ "word_delimiter_token": word_delimiter_token,
588
+ }
589
+
590
+ # 5. Now we can instantiate the feature extractor, tokenizer and model
591
+ # Note for distributed training, the .from_pretrained methods guarantee that only
592
+ # one local process can concurrently download model & vocab.
593
+
594
+ # load feature_extractor and tokenizer
595
+ tokenizer = AutoTokenizer.from_pretrained(
596
+ tokenizer_name_or_path,
597
+ use_auth_token=data_args.use_auth_token,
598
+ **tokenizer_kwargs,
599
+ )
600
+ feature_extractor = AutoFeatureExtractor.from_pretrained(
601
+ model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_auth_token=data_args.use_auth_token
602
+ )
603
+
604
+ # adapt config
605
+ config.update(
606
+ {
607
+ "feat_proj_dropout": model_args.feat_proj_dropout,
608
+ "attention_dropout": model_args.attention_dropout,
609
+ "hidden_dropout": model_args.hidden_dropout,
610
+ "final_dropout": model_args.final_dropout,
611
+ "mask_time_prob": model_args.mask_time_prob,
612
+ "mask_time_length": model_args.mask_time_length,
613
+ "mask_feature_prob": model_args.mask_feature_prob,
614
+ "mask_feature_length": model_args.mask_feature_length,
615
+ "gradient_checkpointing": training_args.gradient_checkpointing,
616
+ "layerdrop": model_args.layerdrop,
617
+ "ctc_loss_reduction": model_args.ctc_loss_reduction,
618
+ "ctc_zero_infinity": model_args.ctc_zero_infinity,
619
+ "pad_token_id": tokenizer.pad_token_id,
620
+ "vocab_size": len(tokenizer),
621
+ "activation_dropout": model_args.activation_dropout,
622
+ }
623
+ )
624
+
625
+ # create model
626
+ model = AutoModelForCTC.from_pretrained(
627
+ model_args.model_name_or_path,
628
+ cache_dir=model_args.cache_dir,
629
+ config=config,
630
+ use_auth_token=data_args.use_auth_token,
631
+ )
632
+
633
+ # freeze encoder
634
+ if model_args.freeze_feature_encoder:
635
+ model.freeze_feature_encoder()
636
+
637
+ # 6. Now we preprocess the datasets including loading the audio, resampling and normalization
638
+ # Thankfully, `datasets` takes care of automatically loading and resampling the audio,
639
+ # so that we just need to set the correct target sampling rate and normalize the input
640
+ # via the `feature_extractor`
641
+
642
+ # make sure that dataset decodes audio with correct sampling rate
643
+ dataset_sampling_rate = next(iter(raw_datasets.values())).features[data_args.audio_column_name].sampling_rate
644
+ if dataset_sampling_rate != feature_extractor.sampling_rate:
645
+ raw_datasets = raw_datasets.cast_column(
646
+ data_args.audio_column_name, datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate)
647
+ )
648
+
649
+ # derive max & min input length for sample rate & max duration
650
+ max_input_length = data_args.max_duration_in_seconds * feature_extractor.sampling_rate
651
+ min_input_length = data_args.min_duration_in_seconds * feature_extractor.sampling_rate
652
+ audio_column_name = data_args.audio_column_name
653
+ num_workers = data_args.preprocessing_num_workers
654
+
655
+ # `phoneme_language` is only relevant if the model is fine-tuned on phoneme classification
656
+ phoneme_language = data_args.phoneme_language
657
+
658
+ # Preprocessing the datasets.
659
+ # We need to read the audio files as arrays and tokenize the targets.
660
+ def prepare_dataset(batch):
661
+ # load audio
662
+ sample = batch[audio_column_name]
663
+
664
+ inputs = feature_extractor(sample["array"], sampling_rate=sample["sampling_rate"])
665
+ batch["input_values"] = inputs.input_values[0]
666
+ batch["input_length"] = len(batch["input_values"])
667
+
668
+ # encode targets
669
+ additional_kwargs = {}
670
+ if phoneme_language is not None:
671
+ additional_kwargs["phonemizer_lang"] = phoneme_language
672
+
673
+ batch["labels"] = tokenizer(batch["target_text"], **additional_kwargs).input_ids
674
+ return batch
675
+
676
+ with training_args.main_process_first(desc="dataset map preprocessing"):
677
+ vectorized_datasets = raw_datasets.map(
678
+ prepare_dataset,
679
+ remove_columns=next(iter(raw_datasets.values())).column_names,
680
+ num_proc=num_workers,
681
+ desc="preprocess datasets",
682
+ )
683
+
684
+ def is_audio_in_length_range(length):
685
+ return length > min_input_length and length < max_input_length
686
+
687
+ # filter data that is shorter than min_input_length
688
+ vectorized_datasets = vectorized_datasets.filter(
689
+ is_audio_in_length_range,
690
+ num_proc=num_workers,
691
+ input_columns=["input_length"],
692
+ )
693
+
694
+ # 7. Next, we can prepare the training.
695
+ # Let's use word error rate (WER) as our evaluation metric,
696
+ # instantiate a data collator and the trainer
697
+
698
+ # Define evaluation metrics during training, *i.e.* word error rate, character error rate
699
+ eval_metrics = {metric: load_metric(metric) for metric in data_args.eval_metrics}
700
+
701
+ # for large datasets it is advised to run the preprocessing on a
702
+ # single machine first with ``args.preprocessing_only`` since there will mostly likely
703
+ # be a timeout when running the script in distributed mode.
704
+ # In a second step ``args.preprocessing_only`` can then be set to `False` to load the
705
+ # cached dataset
706
+ if data_args.preprocessing_only:
707
+ logger.info(f"Data preprocessing finished. Files cached at {vectorized_datasets.cache_files}")
708
+ return
709
+
710
+ def compute_metrics(pred):
711
+ pred_logits = pred.predictions
712
+ pred_ids = np.argmax(pred_logits, axis=-1)
713
+
714
+ pred.label_ids[pred.label_ids == -100] = tokenizer.pad_token_id
715
+
716
+ pred_str = tokenizer.batch_decode(pred_ids)
717
+ # we do not want to group tokens when computing the metrics
718
+ label_str = tokenizer.batch_decode(pred.label_ids, group_tokens=False)
719
+
720
+ metrics = {k: v.compute(predictions=pred_str, references=label_str) for k, v in eval_metrics.items()}
721
+
722
+ return metrics
723
+
724
+ # Now save everything to be able to create a single processor later
725
+ if is_main_process(training_args.local_rank):
726
+ # save feature extractor, tokenizer and config
727
+ feature_extractor.save_pretrained(training_args.output_dir)
728
+ tokenizer.save_pretrained(training_args.output_dir)
729
+ config.save_pretrained(training_args.output_dir)
730
+
731
+ try:
732
+ processor = AutoProcessor.from_pretrained(training_args.output_dir)
733
+ except (OSError, KeyError):
734
+ warnings.warn(
735
+ "Loading a processor from a feature extractor config that does not"
736
+ " include a `processor_class` attribute is deprecated and will be removed in v5. Please add the following "
737
+ " attribute to your `preprocessor_config.json` file to suppress this warning: "
738
+ " `'processor_class': 'Wav2Vec2Processor'`",
739
+ FutureWarning,
740
+ )
741
+ processor = Wav2Vec2Processor.from_pretrained(training_args.output_dir)
742
+
743
+ # Instantiate custom data collator
744
+ data_collator = DataCollatorCTCWithPadding(processor=processor)
745
+
746
+ # Initialize Trainer
747
+ trainer = Trainer(
748
+ model=model,
749
+ data_collator=data_collator,
750
+ args=training_args,
751
+ compute_metrics=compute_metrics,
752
+ train_dataset=vectorized_datasets["train"] if training_args.do_train else None,
753
+ eval_dataset=vectorized_datasets["validation"] if training_args.do_eval else None,
754
+ tokenizer=feature_extractor,
755
+ )
756
+
757
+ # 8. Finally, we can start training
758
+
759
+ # Training
760
+ if training_args.do_train:
761
+
762
+ # use last checkpoint if exist
763
+ if last_checkpoint is not None:
764
+ checkpoint = last_checkpoint
765
+ elif os.path.isdir(model_args.model_name_or_path):
766
+ checkpoint = model_args.model_name_or_path
767
+ else:
768
+ checkpoint = None
769
+
770
+ train_result = trainer.train(resume_from_checkpoint=checkpoint)
771
+ trainer.save_model()
772
+
773
+ metrics = train_result.metrics
774
+ max_train_samples = (
775
+ data_args.max_train_samples
776
+ if data_args.max_train_samples is not None
777
+ else len(vectorized_datasets["train"])
778
+ )
779
+ metrics["train_samples"] = min(max_train_samples, len(vectorized_datasets["train"]))
780
+
781
+ trainer.log_metrics("train", metrics)
782
+ trainer.save_metrics("train", metrics)
783
+ trainer.save_state()
784
+
785
+ # Evaluation
786
+ results = {}
787
+ if training_args.do_eval:
788
+ logger.info("*** Evaluate ***")
789
+ metrics = trainer.evaluate()
790
+ max_eval_samples = (
791
+ data_args.max_eval_samples if data_args.max_eval_samples is not None else len(vectorized_datasets["eval"])
792
+ )
793
+ metrics["eval_samples"] = min(max_eval_samples, len(vectorized_datasets["eval"]))
794
+
795
+ trainer.log_metrics("eval", metrics)
796
+ trainer.save_metrics("eval", metrics)
797
+
798
+ # Write model card and (optionally) push to hub
799
+ config_name = data_args.dataset_config_name if data_args.dataset_config_name is not None else "na"
800
+ kwargs = {
801
+ "finetuned_from": model_args.model_name_or_path,
802
+ "tasks": "speech-recognition",
803
+ "tags": ["automatic-speech-recognition", data_args.dataset_name],
804
+ "dataset_args": f"Config: {config_name}, Training split: {data_args.train_split_name}, Eval split: {data_args.eval_split_name}",
805
+ "dataset": f"{data_args.dataset_name.upper()} - {config_name.upper()}",
806
+ }
807
+ if "common_voice" in data_args.dataset_name:
808
+ kwargs["language"] = config_name
809
+
810
+ if training_args.push_to_hub:
811
+ trainer.push_to_hub(**kwargs)
812
+ else:
813
+ trainer.create_model_card(**kwargs)
814
+
815
+ return results
816
+
817
+
818
+ if __name__ == "__main__":
819
+ main()
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]", "additional_special_tokens": [{"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}]}
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "replace_word_delimiter_char": " ", "special_tokens_map_file": null, "name_or_path": "./", "tokenizer_class": "Wav2Vec2CTCTokenizer"}
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:23b4bcfd09d8da01d82d5123ed1f20b17a5cbd32a53ff26e65918f2c953ba6b7
3
+ size 3055
vocab.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"a": 1, "b": 2, "c": 3, "d": 4, "e": 5, "f": 6, "g": 7, "h": 8, "i": 9, "j": 10, "k": 11, "l": 12, "m": 13, "n": 14, "o": 15, "p": 16, "q": 17, "r": 18, "s": 19, "t": 20, "u": 21, "v": 22, "w": 23, "x": 24, "y": 25, "z": 26, "å": 27, "æ": 28, "ø": 29, "|": 0, "[UNK]": 30, "[PAD]": 31}
wandb/debug-internal.log ADDED
@@ -0,0 +1 @@
 
 
1
+ run-20220523_115145-3dybzmyz/logs/debug-internal.log
wandb/debug.log ADDED
@@ -0,0 +1 @@
 
 
1
+ run-20220523_115145-3dybzmyz/logs/debug.log
wandb/latest-run ADDED
@@ -0,0 +1 @@
 
 
1
+ run-20220523_115145-3dybzmyz
wandb/run-20220523_091609-1iboydmy/files/config.yaml ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_091609-1iboydmy/files/output.log ADDED
@@ -0,0 +1,1379 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+
3
+
4
+
5
+
6
+
7
+
8
+
9
+
10
+
11
+
12
+
13
+
14
+
15
+
16
+
17
+
18
+
19
+
20
+
21
+
22
+
23
+
24
+
25
+
26
+
27
+
28
+
29
+
30
+
31
+
32
+
33
+
34
+
35
+
36
+
37
+
38
+
39
+
40
+
41
+
42
+
43
+
44
+
45
+
46
+
47
+
48
+
49
+
50
+
51
+
52
+
53
+
54
+
55
+
56
+
57
+
58
+
59
+
60
+
61
+
62
+
63
+
64
+
65
+
66
+
67
+
68
+
69
+
70
+
71
+
72
+
73
+
74
+
75
+
76
+
77
+
78
+
79
+
80
+
81
+
82
+
83
+
84
+
85
+
86
+ 0%| | 100/503920 [03:45<123:27:02, 1.13it/s]
87
+
88
+
89
+
90
+
91
+
92
+
93
+
94
+
95
+
96
+
97
+
98
+
99
+
100
+
101
+
102
+
103
+
104
+
105
+
106
+
107
+
108
+
109
+
110
+
111
+
112
+
113
+
114
+
115
+
116
+
117
+
118
+
119
+
120
+
121
+
122
+
123
+
124
+
125
+
126
+
127
+
128
+
129
+
130
+
131
+
132
+
133
+
134
+
135
+
136
+
137
+
138
+
139
+
140
+
141
+
142
+
143
+
144
+
145
+
146
+
147
+
148
+
149
+
150
+
151
+
152
+
153
+
154
+
155
+
156
+
157
+
158
+
159
+
160
+
161
+
162
+
163
+
164
+
165
+
166
+
167
+
168
+
169
+
170
+ 0%| | 199/503920 [07:33<131:30:18, 1.06it/s]
171
+
172
+
173
+
174
+
175
+
176
+
177
+
178
+
179
+
180
+
181
+
182
+
183
+
184
+
185
+
186
+
187
+
188
+
189
+
190
+
191
+
192
+
193
+
194
+
195
+
196
+
197
+
198
+
199
+
200
+
201
+
202
+
203
+
204
+
205
+
206
+
207
+
208
+
209
+
210
+
211
+
212
+
213
+
214
+
215
+
216
+
217
+
218
+
219
+
220
+
221
+
222
+
223
+
224
+
225
+
226
+
227
+
228
+
229
+
230
+
231
+
232
+
233
+
234
+
235
+
236
+
237
+
238
+
239
+
240
+
241
+
242
+
243
+
244
+
245
+
246
+
247
+
248
+
249
+
250
+
251
+
252
+
253
+
254
+
255
+
256
+ 0%| | 298/503920 [11:21<140:58:24, 1.01s/it]
257
+
258
+
259
+
260
+
261
+
262
+
263
+
264
+
265
+
266
+
267
+
268
+
269
+
270
+
271
+
272
+
273
+
274
+
275
+
276
+
277
+
278
+
279
+
280
+
281
+
282
+
283
+
284
+
285
+
286
+
287
+
288
+
289
+
290
+
291
+
292
+
293
+
294
+
295
+
296
+
297
+
298
+
299
+
300
+
301
+
302
+
303
+
304
+
305
+
306
+
307
+
308
+
309
+
310
+
311
+
312
+
313
+
314
+
315
+
316
+
317
+
318
+
319
+
320
+
321
+
322
+
323
+
324
+
325
+
326
+
327
+
328
+
329
+
330
+
331
+
332
+
333
+
334
+
335
+
336
+
337
+
338
+
339
+
340
+
341
+
342
+ 0%| | 400/503920 [15:10<122:42:36, 1.14it/s]
343
+
344
+
345
+
346
+
347
+
348
+
349
+
350
+
351
+
352
+
353
+
354
+
355
+
356
+
357
+
358
+
359
+
360
+
361
+
362
+
363
+
364
+
365
+
366
+
367
+
368
+
369
+
370
+
371
+
372
+
373
+
374
+
375
+
376
+
377
+
378
+
379
+
380
+
381
+
382
+
383
+
384
+
385
+
386
+
387
+
388
+
389
+
390
+
391
+
392
+
393
+
394
+
395
+
396
+
397
+
398
+
399
+
400
+
401
+
402
+
403
+
404
+
405
+
406
+
407
+
408
+
409
+
410
+
411
+
412
+
413
+
414
+
415
+
416
+
417
+
418
+
419
+
420
+
421
+
422
+
423
+
424
+
425
+ 0%|▏ | 500/503920 [18:58<122:35:54, 1.14it/s]The following columns in the evaluation set don't have a corresponding argument in `Wav2Vec2ForCTC.forward` and have been ignored: input_length. If input_length are not expected by `Wav2Vec2ForCTC.forward`, you can safely ignore this message.
426
+ ***** Running Evaluation *****
427
+ Num examples = 41040
428
+ Batch size = 12
429
+ 0%| | 0/3420 [00:00<?, ?it/s]
430
+
431
+
432
+
433
+
434
+
435
+
436
+
437
+
438
+
439
+
440
+
441
+
442
+
443
+
444
+
445
+
446
+
447
+
448
+
449
+
450
+
451
+
452
+
453
+
454
+
455
+
456
+
457
+
458
+
459
+
460
+
461
+
462
+
463
+
464
+
465
+
466
+
467
+
468
+
469
+
470
+
471
+
472
+
473
+
474
+
475
+
476
+
477
+
478
+
479
+
480
+
481
+
482
+
483
+
484
+
485
+
486
+
487
+
488
+
489
+
490
+
491
+
492
+
493
+
494
+
495
+
496
+
497
+
498
+
499
+
500
+
501
+
502
+
503
+
504
+
505
+
506
+
507
+
508
+
509
+
510
+
511
+
512
+
513
+
514
+
515
+
516
+
517
+
518
+
519
+
520
+
521
+
522
+
523
+
524
+
525
+
526
+
527
+
528
+
529
+
530
+
531
+
532
+
533
+
534
+
535
+
536
+
537
+
538
+
539
+
540
+
541
+
542
+
543
+
544
+
545
+
546
+
547
+
548
+
549
+
550
+
551
+
552
+
553
+
554
+
555
+
556
+
557
+
558
+
559
+
560
+
561
+
562
+
563
+
564
+
565
+
566
+
567
+
568
+
569
+
570
+
571
+
572
+
573
+
574
+
575
+
576
+
577
+
578
+
579
+
580
+
581
+
582
+
583
+
584
+
585
+
586
+
587
+
588
+
589
+
590
+
591
+
592
+
593
+
594
+
595
+
596
+
597
+
598
+
599
+
600
+
601
+
602
+
603
+
604
+
605
+
606
+
607
+
608
+
609
+
610
+
611
+
612
+
613
+
614
+
615
+
616
+
617
+
618
+
619
+
620
+
621
+
622
+
623
+
624
+
625
+
626
+
627
+
628
+
629
+
630
+
631
+
632
+
633
+
634
+
635
+
636
+
637
+
638
+
639
+
640
+
641
+
642
+
643
+
644
+
645
+
646
+
647
+
648
+
649
+
650
+
651
+
652
+
653
+
654
+
655
+
656
+
657
+
658
+
659
+
660
+
661
+
662
+
663
+
664
+
665
+
666
+
667
+
668
+
669
+
670
+
671
+
672
+
673
+
674
+
675
+
676
+
677
+
678
+
679
+
680
+
681
+
682
+
683
+
684
+
685
+
686
+
687
+
688
+
689
+
690
+
691
+
692
+
693
+
694
+
695
+
696
+
697
+
698
+
699
+
700
+
701
+
702
+
703
+
704
+
705
+
706
+
707
+
708
+
709
+
710
+
711
+
712
+
713
+
714
+
715
+
716
+
717
+
718
+
719
+
720
+
721
+
722
+
723
+
724
+
725
+
726
+
727
+
728
+
729
+
730
+
731
+
732
+
733
+
734
+
735
+
736
+
737
+
738
+
739
+
740
+
741
+
742
+
743
+
744
+
745
+
746
+
747
+
748
+
749
+
750
+
751
+
752
+
753
+
754
+
755
+
756
+
757
+
758
+
759
+
760
+
761
+
762
+
763
+
764
+
765
+
766
+
767
+
768
+
769
+
770
+
771
+
772
+
773
+
774
+
775
+
776
+
777
+
778
+
779
+
780
+
781
+
782
+
783
+
784
+
785
+
786
+
787
+
788
+
789
+
790
+
791
+
792
+
793
+
794
+
795
+
796
+
797
+
798
+
799
+
800
+
801
+
802
+
803
+
804
+
805
+
806
+
807
+
808
+
809
+
810
+
811
+
812
+
813
+
814
+
815
+
816
+
817
+
818
+
819
+
820
+
821
+
822
+
823
+
824
+
825
+
826
+
827
+
828
+
829
+
830
+
831
+
832
+
833
+
834
+
835
+
836
+
837
+
838
+
839
+
840
+
841
+
842
+
843
+
844
+
845
+
846
+
847
+
848
+
849
+
850
+
851
+
852
+
853
+
854
+
855
+
856
+
857
+
858
+
859
+
860
+
861
+
862
+
863
+
864
+
865
+
866
+
867
+
868
+
869
+
870
+
871
+
872
+
873
+
874
+
875
+
876
+
877
+
878
+
879
+
880
+
881
+
882
+
883
+
884
+
885
+
886
+
887
+
888
+
889
+
890
+
891
+
892
+
893
+
894
+
895
+
896
+
897
+
898
+
899
+
900
+
901
+
902
+
903
+
904
+
905
+
906
+
907
+
908
+
909
+
910
+
911
+
912
+
913
+
914
+
915
+
916
+
917
+
918
+
919
+
920
+
921
+
922
+
923
+
924
+
925
+
926
+
927
+
928
+
929
+
930
+
931
+
932
+
933
+
934
+
935
+
936
+
937
+
938
+
939
+
940
+
941
+
942
+
943
+
944
+
945
+
946
+
947
+
948
+
949
+
950
+
951
+
952
+
953
+
954
+
955
+
956
+
957
+
958
+
959
+
960
+
961
+
962
+
963
+
964
+
965
+
966
+
967
+
968
+
969
+
970
+
971
+
972
+
973
+
974
+
975
+
976
+
977
+
978
+
979
+
980
+
981
+
982
+
983
+
984
+
985
+
986
+
987
+
988
+
989
+
990
+
991
+
992
+
993
+
994
+
995
+
996
+
997
+
998
+
999
+
1000
+
1001
+
1002
+
1003
+
1004
+
1005
+
1006
+
1007
+
1008
+
1009
+
1010
+
1011
+
1012
+
1013
+
1014
+
1015
+
1016
+
1017
+
1018
+
1019
+
1020
+
1021
+
1022
+
1023
+
1024
+
1025
+
1026
+
1027
+
1028
+
1029
+
1030
+
1031
+
1032
+
1033
+
1034
+
1035
+
1036
+
1037
+
1038
+
1039
+
1040
+
1041
+
1042
+
1043
+
1044
+
1045
+
1046
+
1047
+
1048
+
1049
+
1050
+
1051
+
1052
+
1053
+
1054
+
1055
+
1056
+
1057
+
1058
+
1059
+
1060
+
1061
+
1062
+
1063
+
1064
+
1065
+
1066
+
1067
+
1068
+
1069
+
1070
+
1071
+
1072
+
1073
+
1074
+
1075
+
1076
+
1077
+
1078
+
1079
+
1080
+
1081
+
1082
+
1083
+
1084
+
1085
+
1086
+
1087
+
1088
+
1089
+
1090
+
1091
+
1092
+
1093
+
1094
+
1095
+
1096
+
1097
+
1098
+
1099
+
1100
+
1101
+
1102
+
1103
+
1104
+
1105
+
1106
+
1107
+
1108
+
1109
+
1110
+
1111
+
1112
+
1113
+
1114
+
1115
+
1116
+
1117
+
1118
+
1119
+
1120
+
1121
+
1122
+
1123
+
1124
+
1125
+
1126
+
1127
+
1128
+
1129
+
1130
+
1131
+
1132
+
1133
+
1134
+
1135
+
1136
+
1137
+
1138
+
1139
+
1140
+
1141
+
1142
+
1143
+
1144
+
1145
+
1146
+
1147
+
1148
+
1149
+
1150
+
1151
+
1152
+
1153
+
1154
+
1155
+
1156
+
1157
+
1158
+
1159
+
1160
+
1161
+
1162
+
1163
+
1164
+
1165
+
1166
+
1167
+
1168
+
1169
+
1170
+
1171
+
1172
+
1173
+
1174
+
1175
+
1176
+
1177
+
1178
+
1179
+
1180
+
1181
+
1182
+
1183
+
1184
+
1185
+
1186
+
1187
+
1188
+
1189
+
1190
+
1191
+
1192
+
1193
+
1194
+
1195
+
1196
+
1197
+
1198
+
1199
+
1200
+
1201
+
1202
+
1203
+
1204
+
1205
+
1206
+
1207
+
1208
+
1209
+
1210
+
1211
+
1212
+
1213
+
1214
+
1215
+
1216
+
1217
+
1218
+
1219
+
1220
+
1221
+
1222
+
1223
+
1224
+
1225
+
1226
+
1227
+
1228
+
1229
+
1230
+
1231
+
1232
+
1233
+
1234
+
1235
+
1236
+
1237
+
1238
+
1239
+
1240
+
1241
+
1242
+
1243
+
1244
+
1245
+
1246
+
1247
+
1248
+
1249
+
1250
+
1251
+
1252
+
1253
+
1254
+
1255
+
1256
+
1257
+
1258
+
1259
+
1260
+
1261
+
1262
+
1263
+
1264
+
1265
+
1266
+
1267
+
1268
+
1269
+
1270
+
1271
+
1272
+
1273
+
1274
+
1275
+
1276
+
1277
+
1278
+
1279
+
1280
+
1281
+
1282
+
1283
+
1284
+
1285
+
1286
+
1287
+
1288
+
1289
+
1290
+
1291
+
1292
+
1293
+
1294
+
1295
+
1296
+
1297
+
1298
+
1299
+
1300
+
1301
+
1302
+
1303
+
1304
+
1305
+
1306
+
1307
+
1308
+
1309
+
1310
+
1311
+
1312
+
1313
+
1314
+
1315
+
1316
+
1317
+
1318
+
1319
+
1320
+
1321
+
1322
+
1323
+
1324
+
1325
+
1326
+
1327
+
1328
+
1329
+
1330
+
1331
+
1332
+
1333
+
1334
+
1335
+
1336
+
1337
+
1338
+
1339
+ File "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py", line 819, in <module> | 2355/3420 [30:44<12:33, 1.41it/s]
1340
+ main()
1341
+ File "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py", line 770, in main
1342
+ train_result = trainer.train(resume_from_checkpoint=checkpoint)
1343
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 1497, in train
1344
+ self._maybe_log_save_evaluate(tr_loss, model, trial, epoch, ignore_keys_for_eval)
1345
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 1624, in _maybe_log_save_evaluate
1346
+ metrics = self.evaluate(ignore_keys=ignore_keys_for_eval)
1347
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 2284, in evaluate
1348
+ output = eval_loop(
1349
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 2458, in evaluation_loop
1350
+ loss, logits, labels = self.prediction_step(model, inputs, prediction_loss_only, ignore_keys=ignore_keys)
1351
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 2671, in prediction_step
1352
+ loss, outputs = self.compute_loss(model, inputs, return_outputs=True)
1353
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 2043, in compute_loss
1354
+ outputs = model(**inputs)
1355
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
1356
+ return forward_call(*input, **kwargs)
1357
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/models/wav2vec2/modeling_wav2vec2.py", line 1716, in forward
1358
+ outputs = self.wav2vec2(
1359
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
1360
+ return forward_call(*input, **kwargs)
1361
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/models/wav2vec2/modeling_wav2vec2.py", line 1361, in forward
1362
+ encoder_outputs = self.encoder(
1363
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
1364
+ return forward_call(*input, **kwargs)
1365
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/models/wav2vec2/modeling_wav2vec2.py", line 930, in forward
1366
+ layer_outputs = layer(
1367
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
1368
+ return forward_call(*input, **kwargs)
1369
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/models/wav2vec2/modeling_wav2vec2.py", line 766, in forward
1370
+ hidden_states, attn_weights, _ = self.attention(
1371
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
1372
+ return forward_call(*input, **kwargs)
1373
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/models/wav2vec2/modeling_wav2vec2.py", line 686, in forward
1374
+ attn_output = self.out_proj(attn_output)
1375
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
1376
+ return forward_call(*input, **kwargs)
1377
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/nn/modules/linear.py", line 103, in forward
1378
+ return F.linear(input, self.weight, self.bias)
1379
+ KeyboardInterrupt
wandb/run-20220523_091609-1iboydmy/files/requirements.txt ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ aiohttp==3.8.1
2
+ aiosignal==1.2.0
3
+ appdirs==1.4.4
4
+ async-timeout==4.0.2
5
+ attrs==21.4.0
6
+ audioread==2.1.9
7
+ certifi==2021.10.8
8
+ cffi==1.15.0
9
+ charset-normalizer==2.0.12
10
+ click==8.1.2
11
+ datasets==2.1.0
12
+ decorator==5.1.1
13
+ dill==0.3.4
14
+ docker-pycreds==0.4.0
15
+ filelock==3.6.0
16
+ frozenlist==1.3.0
17
+ fsspec==2022.3.0
18
+ gitdb==4.0.9
19
+ gitpython==3.1.27
20
+ huggingface-hub==0.5.1
21
+ hypothesis==6.46.5
22
+ idna==3.3
23
+ jiwer==2.3.0
24
+ joblib==1.1.0
25
+ kenlm==0.0.0
26
+ librosa==0.9.1
27
+ llvmlite==0.38.0
28
+ multidict==6.0.2
29
+ multiprocess==0.70.12.2
30
+ numba==0.55.1
31
+ numpy==1.21.6
32
+ packaging==21.3
33
+ pandas==1.4.2
34
+ pathtools==0.1.2
35
+ pillow==9.1.0
36
+ pip==20.3.4
37
+ pkg-resources==0.0.0
38
+ pooch==1.6.0
39
+ promise==2.3
40
+ protobuf==3.20.1
41
+ psutil==5.9.0
42
+ pyarrow==7.0.0
43
+ pycparser==2.21
44
+ pyctcdecode==0.3.0
45
+ pygtrie==2.4.2
46
+ pyparsing==3.0.8
47
+ python-dateutil==2.8.2
48
+ python-levenshtein==0.12.2
49
+ pytz==2022.1
50
+ pyyaml==6.0
51
+ regex==2022.4.24
52
+ requests==2.27.1
53
+ resampy==0.2.2
54
+ responses==0.18.0
55
+ sacremoses==0.0.49
56
+ scikit-learn==1.0.2
57
+ scipy==1.8.0
58
+ sentry-sdk==1.5.10
59
+ setproctitle==1.2.3
60
+ setuptools==44.1.1
61
+ shortuuid==1.0.8
62
+ six==1.16.0
63
+ smmap==5.0.0
64
+ sortedcontainers==2.4.0
65
+ soundfile==0.10.3.post1
66
+ threadpoolctl==3.1.0
67
+ tokenizers==0.12.1
68
+ torch==1.11.0+cu113
69
+ torchaudio==0.11.0+cu113
70
+ torchvision==0.12.0+cu113
71
+ tqdm==4.64.0
72
+ transformers==4.18.0
73
+ typing-extensions==4.2.0
74
+ urllib3==1.26.9
75
+ wandb==0.12.15
76
+ xxhash==3.0.0
77
+ yarl==1.7.2
wandb/run-20220523_091609-1iboydmy/files/wandb-metadata.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "os": "Linux-5.13.0-40-generic-x86_64-with-glibc2.34",
3
+ "python": "3.9.7",
4
+ "heartbeatAt": "2022-05-23T07:16:10.988629",
5
+ "startedAt": "2022-05-23T07:16:09.920284",
6
+ "docker": null,
7
+ "cpu_count": 96,
8
+ "cuda": null,
9
+ "args": [
10
+ "--model_name_or_path=facebook/wav2vec2-xls-r-1b",
11
+ "--hub_model_id=NbAiLab/wav2vec2-1b-npsc-nst-bokmaal",
12
+ "--output_dir=./",
13
+ "--overwrite_output_dir",
14
+ "--num_train_epochs=40",
15
+ "--per_device_train_batch_size=12",
16
+ "--per_device_eval_batch_size=12",
17
+ "--gradient_accumulation_steps=2",
18
+ "--learning_rate=2e-5",
19
+ "--warmup_steps=2000",
20
+ "--length_column_name=input_length",
21
+ "--evaluation_strategy=steps",
22
+ "--text_column_name=text",
23
+ "--save_steps=500",
24
+ "--eval_steps=500",
25
+ "--logging_steps=100",
26
+ "--layerdrop=0.041",
27
+ "--attention_dropout=0.094",
28
+ "--activation_dropout=0.055",
29
+ "--hidden_dropout=0.047",
30
+ "--save_total_limit=3",
31
+ "--freeze_feature_encoder",
32
+ "--feat_proj_dropout=0.04",
33
+ "--mask_time_prob=0.082",
34
+ "--mask_time_length=10",
35
+ "--mask_feature_prob=0.25",
36
+ "--mask_feature_length=64",
37
+ "--gradient_checkpointing",
38
+ "--min_duration_in_seconds=0.5",
39
+ "--max_duration_in_seconds=30.0",
40
+ "--use_auth_token",
41
+ "--seed=42",
42
+ "--fp16",
43
+ "--group_by_length",
44
+ "--do_train",
45
+ "--do_eval",
46
+ "--push_to_hub",
47
+ "--preprocessing_num_workers=32",
48
+ "--ctc_zero_infinity"
49
+ ],
50
+ "state": "running",
51
+ "program": "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py",
52
+ "codePath": "run_speech_recognition_ctc.py",
53
+ "git": {
54
+ "remote": "https://huggingface.co/NbAiLab/wav2vec2-1b-npsc-nst-bokmaal",
55
+ "commit": "05ba5dd22c9cec7365fba75fae8630eec1302aff"
56
+ },
57
+ "email": "[email protected]",
58
+ "root": "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal",
59
+ "host": "dante",
60
+ "username": "rolvb",
61
+ "executable": "/mnt/lv_ai_1_dante/ml/rolvb/venv/bin/python"
62
+ }
wandb/run-20220523_091609-1iboydmy/files/wandb-summary.json ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_091609-1iboydmy/logs/debug-internal.log ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_091609-1iboydmy/logs/debug.log ADDED
@@ -0,0 +1,189 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2022-05-23 09:16:09,931 INFO MainThread:1046672 [wandb_setup.py:_flush():75] Loading settings from /home/rolvb/.config/wandb/settings
2
+ 2022-05-23 09:16:09,931 INFO MainThread:1046672 [wandb_setup.py:_flush():75] Loading settings from /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/settings
3
+ 2022-05-23 09:16:09,931 INFO MainThread:1046672 [wandb_setup.py:_flush():75] Loading settings from environment variables: {'project': 'wav2vec2', 'entity': 'NbAiLab'}
4
+ 2022-05-23 09:16:09,931 INFO MainThread:1046672 [wandb_setup.py:_flush():75] Inferring run settings from compute environment: {'program_relpath': 'run_speech_recognition_ctc.py', 'program': '/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py'}
5
+ 2022-05-23 09:16:09,931 INFO MainThread:1046672 [wandb_init.py:_log_setup():437] Logging user logs to /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/run-20220523_091609-1iboydmy/logs/debug.log
6
+ 2022-05-23 09:16:09,931 INFO MainThread:1046672 [wandb_init.py:_log_setup():438] Logging internal logs to /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/run-20220523_091609-1iboydmy/logs/debug-internal.log
7
+ 2022-05-23 09:16:09,932 INFO MainThread:1046672 [wandb_init.py:init():471] calling init triggers
8
+ 2022-05-23 09:16:09,932 INFO MainThread:1046672 [wandb_init.py:init():474] wandb.init called with sweep_config: {}
9
+ config: {}
10
+ 2022-05-23 09:16:09,932 INFO MainThread:1046672 [wandb_init.py:init():524] starting backend
11
+ 2022-05-23 09:16:09,932 INFO MainThread:1046672 [backend.py:_multiprocessing_setup():97] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
12
+ 2022-05-23 09:16:10,011 INFO MainThread:1046672 [backend.py:ensure_launched():217] starting backend process...
13
+ 2022-05-23 09:16:10,086 INFO MainThread:1046672 [backend.py:ensure_launched():222] started backend process with pid: 1047416
14
+ 2022-05-23 09:16:10,088 INFO MainThread:1046672 [wandb_init.py:init():533] backend started and connected
15
+ 2022-05-23 09:16:10,099 INFO MainThread:1046672 [wandb_init.py:init():597] updated telemetry
16
+ 2022-05-23 09:16:10,290 INFO MainThread:1046672 [wandb_init.py:init():628] communicating run to backend with 30 second timeout
17
+ 2022-05-23 09:16:10,843 INFO MainThread:1046672 [wandb_run.py:_on_init():1923] communicating current version
18
+ 2022-05-23 09:16:10,973 INFO MainThread:1046672 [wandb_run.py:_on_init():1927] got version response upgrade_message: "wandb version 0.12.16 is available! To upgrade, please run:\n $ pip install wandb --upgrade"
19
+
20
+ 2022-05-23 09:16:10,973 INFO MainThread:1046672 [wandb_init.py:init():659] starting run threads in backend
21
+ 2022-05-23 09:16:11,023 INFO MainThread:1046672 [wandb_run.py:_console_start():1897] atexit reg
22
+ 2022-05-23 09:16:11,024 INFO MainThread:1046672 [wandb_run.py:_redirect():1770] redirect: SettingsConsole.REDIRECT
23
+ 2022-05-23 09:16:11,024 INFO MainThread:1046672 [wandb_run.py:_redirect():1775] Redirecting console.
24
+ 2022-05-23 09:16:11,027 INFO MainThread:1046672 [wandb_run.py:_redirect():1831] Redirects installed.
25
+ 2022-05-23 09:16:11,027 INFO MainThread:1046672 [wandb_init.py:init():684] run started, returning control to user process
26
+ 2022-05-23 09:16:11,051 INFO MainThread:1046672 [wandb_run.py:_config_callback():1131] config_cb None None {'return_dict': True, 'output_hidden_states': False, 'output_attentions': False, 'torchscript': False, 'torch_dtype': 'float32', 'use_bfloat16': False, 'pruned_heads': {}, 'tie_word_embeddings': True, 'is_encoder_decoder': False, 'is_decoder': False, 'cross_attention_hidden_size': None, 'add_cross_attention': False, 'tie_encoder_decoder': False, 'max_length': 20, 'min_length': 0, 'do_sample': False, 'early_stopping': False, 'num_beams': 1, 'num_beam_groups': 1, 'diversity_penalty': 0.0, 'temperature': 1.0, 'top_k': 50, 'top_p': 1.0, 'typical_p': 1.0, 'repetition_penalty': 1.0, 'length_penalty': 1.0, 'no_repeat_ngram_size': 0, 'encoder_no_repeat_ngram_size': 0, 'bad_words_ids': None, 'num_return_sequences': 1, 'chunk_size_feed_forward': 0, 'output_scores': False, 'return_dict_in_generate': False, 'forced_bos_token_id': None, 'forced_eos_token_id': None, 'remove_invalid_values': False, 'exponential_decay_length_penalty': None, 'architectures': ['Wav2Vec2ForPreTraining'], 'finetuning_task': None, 'id2label': {0: 'LABEL_0', 1: 'LABEL_1'}, 'label2id': {'LABEL_0': 0, 'LABEL_1': 1}, 'tokenizer_class': None, 'prefix': None, 'bos_token_id': 1, 'pad_token_id': 31, 'eos_token_id': 2, 'sep_token_id': None, 'decoder_start_token_id': None, 'task_specific_params': None, 'problem_type': None, '_name_or_path': 'facebook/wav2vec2-xls-r-1b', 'transformers_version': '4.18.0', 'feat_extract_dropout': 0.0, 'model_type': 'wav2vec2', 'num_feat_extract_layers': 7, 'hidden_size': 1280, 'feat_extract_norm': 'layer', 'feat_extract_activation': 'gelu', 'conv_dim': [512, 512, 512, 512, 512, 512, 512], 'conv_stride': [5, 2, 2, 2, 2, 2, 2], 'conv_kernel': [10, 3, 3, 3, 3, 2, 2], 'conv_bias': True, 'num_conv_pos_embeddings': 128, 'num_conv_pos_embedding_groups': 16, 'num_hidden_layers': 48, 'intermediate_size': 5120, 'hidden_act': 'gelu', 'num_attention_heads': 16, 'hidden_dropout': 0.047, 'attention_dropout': 0.094, 'activation_dropout': 0.055, 'feat_proj_dropout': 0.04, 'final_dropout': 0.0, 'layerdrop': 0.041, 'layer_norm_eps': 1e-05, 'initializer_range': 0.02, 'vocab_size': 34, 'do_stable_layer_norm': True, 'use_weighted_layer_sum': False, 'apply_spec_augment': True, 'mask_time_prob': 0.082, 'mask_time_length': 10, 'mask_time_min_masks': 2, 'mask_feature_prob': 0.25, 'mask_feature_length': 64, 'mask_feature_min_masks': 0, 'num_codevectors_per_group': 320, 'num_codevector_groups': 2, 'contrastive_logits_temperature': 0.1, 'feat_quantizer_dropout': 0.0, 'num_negatives': 100, 'codevector_dim': 1024, 'proj_codevector_dim': 1024, 'diversity_loss_weight': 0.1, 'ctc_loss_reduction': 'mean', 'ctc_zero_infinity': True, 'add_adapter': False, 'adapter_kernel_size': 3, 'adapter_stride': 2, 'num_adapter_layers': 3, 'output_hidden_size': 1280, 'classifier_proj_size': 256, 'tdnn_dim': [512, 512, 512, 512, 1500], 'tdnn_kernel': [5, 3, 3, 1, 1], 'tdnn_dilation': [1, 2, 3, 1, 1], 'xvector_output_dim': 512, 'output_dir': './', 'overwrite_output_dir': True, 'do_train': True, 'do_eval': True, 'do_predict': False, 'evaluation_strategy': 'steps', 'prediction_loss_only': False, 'per_device_train_batch_size': 12, 'per_device_eval_batch_size': 12, 'per_gpu_train_batch_size': 'None', 'per_gpu_eval_batch_size': 'None', 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': 'None', 'eval_delay': 0, 'learning_rate': 2e-05, 'weight_decay': 0.0, 'adam_beta1': 0.9, 'adam_beta2': 0.999, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 40.0, 'max_steps': -1, 'lr_scheduler_type': 'linear', 'warmup_ratio': 0.0, 'warmup_steps': 2000, 'log_level': -1, 'log_level_replica': -1, 'log_on_each_node': True, 'logging_dir': './runs/May23_08-59-49_dante', 'logging_strategy': 'steps', 'logging_first_step': False, 'logging_steps': 100, 'logging_nan_inf_filter': True, 'save_strategy': 'steps', 'save_steps': 500, 'save_total_limit': 3, 'save_on_each_node': False, 'no_cuda': False, 'seed': 42, 'data_seed': 'None', 'bf16': False, 'fp16': True, 'fp16_opt_level': 'O1', 'half_precision_backend': 'amp', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': 'None', 'local_rank': -1, 'xpu_backend': 'None', 'tpu_num_cores': 'None', 'tpu_metrics_debug': False, 'debug': '[]', 'dataloader_drop_last': False, 'eval_steps': 500, 'dataloader_num_workers': 0, 'past_index': -1, 'run_name': './', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': 'None', 'load_best_model_at_end': False, 'metric_for_best_model': 'None', 'greater_is_better': 'None', 'ignore_data_skip': False, 'sharded_ddp': '[]', 'deepspeed': 'None', 'label_smoothing_factor': 0.0, 'optim': 'adamw_hf', 'adafactor': False, 'group_by_length': True, 'length_column_name': 'input_length', 'report_to': "['wandb']", 'ddp_find_unused_parameters': 'None', 'ddp_bucket_cap_mb': 'None', 'dataloader_pin_memory': True, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': True, 'resume_from_checkpoint': 'None', 'hub_model_id': 'NbAiLab/wav2vec2-1b-npsc-nst-bokmaal', 'hub_strategy': 'every_save', 'hub_token': '<HUB_TOKEN>', 'gradient_checkpointing': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': 'None', 'push_to_hub_organization': 'None', 'push_to_hub_token': '<PUSH_TO_HUB_TOKEN>', '_n_gpu': 1, 'mp_parameters': '', 'train_batch_size': 12, 'eval_batch_size': 12}
27
+ 2022-05-23 09:16:11,054 INFO MainThread:1046672 [wandb_watch.py:watch():47] Watching
28
+ 2022-05-23 10:05:55,537 INFO MainThread:1046672 [wandb_run.py:_atexit_cleanup():1866] got exitcode: 255
29
+ 2022-05-23 10:05:55,539 INFO MainThread:1046672 [wandb_run.py:_restore():1838] restore
30
+ 2022-05-23 10:05:58,337 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
31
+ wandb_count: 1
32
+ }
33
+ pusher_stats {
34
+ uploaded_bytes: 2176
35
+ total_bytes: 2176
36
+ }
37
+
38
+ 2022-05-23 10:05:58,442 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
39
+ wandb_count: 1
40
+ }
41
+ pusher_stats {
42
+ uploaded_bytes: 2176
43
+ total_bytes: 2176
44
+ }
45
+
46
+ 2022-05-23 10:05:58,584 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
47
+ wandb_count: 1
48
+ }
49
+ pusher_stats {
50
+ uploaded_bytes: 2176
51
+ total_bytes: 2176
52
+ }
53
+
54
+ 2022-05-23 10:05:59,811 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
55
+ wandb_count: 1
56
+ }
57
+ pusher_stats {
58
+ uploaded_bytes: 2176
59
+ total_bytes: 2176
60
+ }
61
+
62
+ 2022-05-23 10:06:00,251 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
63
+ wandb_count: 5
64
+ }
65
+ pusher_stats {
66
+ uploaded_bytes: 2176
67
+ total_bytes: 1784001
68
+ }
69
+
70
+ 2022-05-23 10:06:00,354 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
71
+ wandb_count: 5
72
+ }
73
+ pusher_stats {
74
+ uploaded_bytes: 2176
75
+ total_bytes: 1784001
76
+ }
77
+
78
+ 2022-05-23 10:06:00,456 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
79
+ wandb_count: 5
80
+ }
81
+ pusher_stats {
82
+ uploaded_bytes: 2176
83
+ total_bytes: 1784001
84
+ }
85
+
86
+ 2022-05-23 10:06:00,558 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
87
+ wandb_count: 5
88
+ }
89
+ pusher_stats {
90
+ uploaded_bytes: 2176
91
+ total_bytes: 1784001
92
+ }
93
+
94
+ 2022-05-23 10:06:00,661 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
95
+ wandb_count: 5
96
+ }
97
+ pusher_stats {
98
+ uploaded_bytes: 1012683
99
+ total_bytes: 1784001
100
+ }
101
+
102
+ 2022-05-23 10:06:00,763 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
103
+ wandb_count: 5
104
+ }
105
+ pusher_stats {
106
+ uploaded_bytes: 1784001
107
+ total_bytes: 1784001
108
+ }
109
+
110
+ 2022-05-23 10:06:00,865 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
111
+ wandb_count: 5
112
+ }
113
+ pusher_stats {
114
+ uploaded_bytes: 1784001
115
+ total_bytes: 1784001
116
+ }
117
+
118
+ 2022-05-23 10:06:00,968 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
119
+ wandb_count: 5
120
+ }
121
+ pusher_stats {
122
+ uploaded_bytes: 1784001
123
+ total_bytes: 1784001
124
+ }
125
+
126
+ 2022-05-23 10:06:01,070 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
127
+ wandb_count: 5
128
+ }
129
+ pusher_stats {
130
+ uploaded_bytes: 1784001
131
+ total_bytes: 1784001
132
+ }
133
+
134
+ 2022-05-23 10:06:01,173 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
135
+ wandb_count: 5
136
+ }
137
+ pusher_stats {
138
+ uploaded_bytes: 1784001
139
+ total_bytes: 1784001
140
+ }
141
+
142
+ 2022-05-23 10:06:01,275 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
143
+ wandb_count: 5
144
+ }
145
+ pusher_stats {
146
+ uploaded_bytes: 1784001
147
+ total_bytes: 1784001
148
+ }
149
+
150
+ 2022-05-23 10:06:01,377 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
151
+ wandb_count: 5
152
+ }
153
+ pusher_stats {
154
+ uploaded_bytes: 1784001
155
+ total_bytes: 1784001
156
+ }
157
+
158
+ 2022-05-23 10:06:01,480 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
159
+ wandb_count: 5
160
+ }
161
+ pusher_stats {
162
+ uploaded_bytes: 1784001
163
+ total_bytes: 1784001
164
+ }
165
+
166
+ 2022-05-23 10:06:03,970 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
167
+ wandb_count: 5
168
+ }
169
+ pusher_stats {
170
+ uploaded_bytes: 1784001
171
+ total_bytes: 1784001
172
+ }
173
+
174
+ 2022-05-23 10:06:04,516 INFO MainThread:1046672 [wandb_run.py:_on_finish():1995] got exit ret: done: true
175
+ exit_result {
176
+ }
177
+ file_counts {
178
+ wandb_count: 5
179
+ }
180
+ pusher_stats {
181
+ uploaded_bytes: 1784001
182
+ total_bytes: 1784001
183
+ }
184
+ local_info {
185
+ }
186
+
187
+ 2022-05-23 10:06:05,688 INFO MainThread:1046672 [wandb_run.py:_footer_history_summary_info():3102] rendering history
188
+ 2022-05-23 10:06:05,726 INFO MainThread:1046672 [wandb_run.py:_footer_history_summary_info():3134] rendering summary
189
+ 2022-05-23 10:06:05,728 INFO MainThread:1046672 [wandb_run.py:_footer_sync_info():3057] logging synced files
wandb/run-20220523_091609-1iboydmy/run-1iboydmy.wandb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd7c0bc4fb7eb8fa38c5bf3141b0fd8413e8309baba69f80901daf5b001d3159
3
+ size 9201298
wandb/run-20220523_103002-wygrs7tw/files/config.yaml ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_103002-wygrs7tw/files/output.log ADDED
@@ -0,0 +1,1746 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+
3
+
4
+
5
+
6
+
7
+
8
+
9
+
10
+
11
+
12
+
13
+
14
+
15
+
16
+
17
+
18
+
19
+
20
+
21
+
22
+
23
+
24
+
25
+
26
+
27
+
28
+
29
+
30
+
31
+
32
+
33
+
34
+
35
+
36
+
37
+
38
+
39
+
40
+
41
+
42
+
43
+
44
+
45
+
46
+
47
+
48
+
49
+
50
+
51
+
52
+
53
+
54
+
55
+
56
+
57
+
58
+
59
+
60
+
61
+
62
+
63
+
64
+
65
+
66
+
67
+
68
+
69
+
70
+
71
+
72
+
73
+
74
+
75
+
76
+
77
+
78
+
79
+
80
+
81
+
82
+
83
+
84
+
85
+
86
+
87
+ 0%| | 100/503920 [03:55<127:33:26, 1.10it/s]
88
+
89
+
90
+
91
+
92
+
93
+
94
+
95
+
96
+
97
+
98
+
99
+
100
+
101
+
102
+
103
+
104
+
105
+
106
+
107
+
108
+
109
+
110
+
111
+
112
+
113
+
114
+
115
+
116
+
117
+
118
+
119
+
120
+
121
+
122
+
123
+
124
+
125
+
126
+
127
+
128
+
129
+
130
+
131
+
132
+
133
+
134
+
135
+
136
+
137
+
138
+
139
+
140
+
141
+
142
+
143
+
144
+
145
+
146
+
147
+
148
+
149
+
150
+
151
+
152
+
153
+
154
+
155
+
156
+
157
+
158
+
159
+
160
+
161
+
162
+
163
+
164
+
165
+
166
+
167
+
168
+
169
+
170
+
171
+
172
+ 0%| | 199/503920 [07:55<136:12:40, 1.03it/s]
173
+
174
+
175
+
176
+
177
+
178
+
179
+
180
+
181
+
182
+
183
+
184
+
185
+
186
+
187
+
188
+
189
+
190
+
191
+
192
+
193
+
194
+
195
+
196
+
197
+
198
+
199
+
200
+
201
+
202
+
203
+
204
+
205
+
206
+
207
+
208
+
209
+
210
+
211
+
212
+
213
+
214
+
215
+
216
+
217
+
218
+
219
+
220
+
221
+
222
+
223
+
224
+
225
+
226
+
227
+
228
+
229
+
230
+
231
+
232
+
233
+
234
+
235
+
236
+
237
+
238
+
239
+
240
+
241
+
242
+
243
+
244
+
245
+
246
+
247
+
248
+
249
+
250
+
251
+
252
+
253
+
254
+
255
+
256
+
257
+
258
+
259
+
260
+ 0%| | 299/503920 [11:55<140:30:23, 1.00s/it]
261
+
262
+
263
+
264
+
265
+
266
+
267
+
268
+
269
+
270
+
271
+
272
+
273
+
274
+
275
+
276
+
277
+
278
+
279
+
280
+
281
+
282
+
283
+
284
+
285
+
286
+
287
+
288
+
289
+
290
+
291
+
292
+
293
+
294
+
295
+
296
+
297
+
298
+
299
+
300
+
301
+
302
+
303
+
304
+
305
+
306
+
307
+
308
+
309
+
310
+
311
+
312
+
313
+
314
+
315
+
316
+
317
+
318
+
319
+
320
+
321
+
322
+
323
+
324
+
325
+
326
+
327
+
328
+
329
+
330
+
331
+
332
+
333
+
334
+
335
+
336
+
337
+
338
+
339
+
340
+
341
+
342
+
343
+
344
+
345
+
346
+ 0%| | 398/503920 [15:47<145:22:16, 1.04s/it]
347
+
348
+
349
+
350
+
351
+
352
+
353
+
354
+
355
+
356
+
357
+
358
+
359
+
360
+
361
+
362
+
363
+
364
+
365
+
366
+
367
+
368
+
369
+
370
+
371
+
372
+
373
+
374
+
375
+
376
+
377
+
378
+
379
+
380
+
381
+
382
+
383
+
384
+
385
+
386
+
387
+
388
+
389
+
390
+
391
+
392
+
393
+
394
+
395
+
396
+
397
+
398
+
399
+
400
+
401
+
402
+
403
+
404
+
405
+
406
+
407
+
408
+
409
+
410
+
411
+
412
+
413
+
414
+
415
+
416
+
417
+
418
+
419
+
420
+
421
+
422
+
423
+
424
+
425
+
426
+
427
+
428
+
429
+
430
+
431
+ 0%|▏ | 500/503920 [19:41<127:22:04, 1.10it/s]The following columns in the evaluation set don't have a corresponding argument in `Wav2Vec2ForCTC.forward` and have been ignored: input_length. If input_length are not expected by `Wav2Vec2ForCTC.forward`, you can safely ignore this message.
432
+ ***** Running Evaluation *****
433
+ Num examples = 41040
434
+ Batch size = 12
435
+ {'loss': 2.9263, 'learning_rate': 4.9000000000000005e-06, 'epoch': 0.04}
436
+
437
+
438
+
439
+
440
+
441
+
442
+
443
+
444
+
445
+
446
+
447
+
448
+
449
+
450
+
451
+
452
+
453
+
454
+
455
+
456
+
457
+
458
+
459
+
460
+
461
+
462
+
463
+
464
+
465
+
466
+
467
+
468
+
469
+
470
+
471
+
472
+
473
+
474
+
475
+
476
+
477
+
478
+
479
+
480
+
481
+
482
+
483
+
484
+
485
+
486
+
487
+
488
+
489
+
490
+
491
+
492
+
493
+
494
+
495
+
496
+
497
+
498
+
499
+
500
+
501
+
502
+
503
+
504
+
505
+
506
+
507
+
508
+
509
+
510
+
511
+
512
+
513
+
514
+
515
+
516
+
517
+
518
+
519
+
520
+
521
+
522
+
523
+
524
+
525
+
526
+
527
+
528
+
529
+
530
+
531
+
532
+
533
+
534
+
535
+
536
+
537
+
538
+
539
+
540
+
541
+
542
+
543
+
544
+
545
+
546
+
547
+
548
+
549
+
550
+
551
+
552
+
553
+
554
+
555
+
556
+
557
+
558
+
559
+
560
+
561
+
562
+
563
+
564
+
565
+
566
+
567
+
568
+
569
+
570
+
571
+
572
+
573
+
574
+
575
+
576
+
577
+
578
+
579
+
580
+
581
+
582
+
583
+
584
+
585
+
586
+
587
+
588
+
589
+
590
+
591
+
592
+
593
+
594
+
595
+
596
+
597
+
598
+
599
+
600
+
601
+
602
+
603
+
604
+
605
+
606
+
607
+
608
+
609
+
610
+
611
+
612
+
613
+
614
+
615
+
616
+
617
+
618
+
619
+
620
+
621
+
622
+
623
+
624
+
625
+
626
+
627
+
628
+
629
+
630
+
631
+
632
+
633
+
634
+
635
+
636
+
637
+
638
+
639
+
640
+
641
+
642
+
643
+
644
+
645
+
646
+
647
+
648
+
649
+
650
+
651
+
652
+
653
+
654
+
655
+
656
+
657
+
658
+
659
+
660
+
661
+
662
+
663
+
664
+
665
+
666
+
667
+
668
+
669
+
670
+
671
+
672
+
673
+
674
+
675
+
676
+
677
+
678
+
679
+
680
+
681
+
682
+
683
+
684
+
685
+
686
+
687
+
688
+
689
+
690
+
691
+
692
+
693
+
694
+
695
+
696
+
697
+
698
+
699
+
700
+
701
+
702
+
703
+
704
+
705
+
706
+
707
+
708
+
709
+
710
+
711
+
712
+
713
+
714
+
715
+
716
+
717
+
718
+
719
+
720
+
721
+
722
+
723
+
724
+
725
+
726
+
727
+
728
+
729
+
730
+
731
+
732
+
733
+
734
+
735
+
736
+
737
+
738
+
739
+
740
+
741
+
742
+
743
+
744
+
745
+
746
+
747
+
748
+
749
+
750
+
751
+
752
+
753
+
754
+
755
+
756
+
757
+
758
+
759
+
760
+
761
+
762
+
763
+
764
+
765
+
766
+
767
+
768
+
769
+
770
+
771
+
772
+
773
+
774
+
775
+
776
+
777
+
778
+
779
+
780
+
781
+
782
+
783
+
784
+
785
+
786
+
787
+
788
+
789
+
790
+
791
+
792
+
793
+
794
+
795
+
796
+
797
+
798
+
799
+
800
+
801
+
802
+
803
+
804
+
805
+
806
+
807
+
808
+
809
+
810
+
811
+
812
+
813
+
814
+
815
+
816
+
817
+
818
+
819
+
820
+
821
+
822
+
823
+
824
+
825
+
826
+
827
+
828
+
829
+
830
+
831
+
832
+
833
+
834
+
835
+
836
+
837
+
838
+
839
+
840
+
841
+
842
+
843
+
844
+
845
+
846
+
847
+
848
+
849
+
850
+
851
+
852
+
853
+
854
+
855
+
856
+
857
+
858
+
859
+
860
+
861
+
862
+
863
+
864
+
865
+
866
+
867
+
868
+
869
+
870
+
871
+
872
+
873
+
874
+
875
+
876
+
877
+
878
+
879
+
880
+
881
+
882
+
883
+
884
+
885
+
886
+
887
+
888
+
889
+
890
+
891
+
892
+
893
+
894
+
895
+
896
+
897
+
898
+
899
+
900
+
901
+
902
+
903
+
904
+
905
+
906
+
907
+
908
+
909
+
910
+
911
+
912
+
913
+
914
+
915
+
916
+
917
+
918
+
919
+
920
+
921
+
922
+
923
+
924
+
925
+
926
+
927
+
928
+
929
+
930
+
931
+
932
+
933
+
934
+
935
+
936
+
937
+
938
+
939
+
940
+
941
+
942
+
943
+
944
+
945
+
946
+
947
+
948
+
949
+
950
+
951
+
952
+
953
+
954
+
955
+
956
+
957
+
958
+
959
+
960
+
961
+
962
+
963
+
964
+
965
+
966
+
967
+
968
+
969
+
970
+
971
+
972
+
973
+
974
+
975
+
976
+
977
+
978
+
979
+
980
+
981
+
982
+
983
+
984
+
985
+
986
+
987
+
988
+
989
+
990
+
991
+
992
+
993
+
994
+
995
+
996
+
997
+
998
+
999
+
1000
+
1001
+
1002
+
1003
+
1004
+
1005
+
1006
+
1007
+
1008
+
1009
+
1010
+
1011
+
1012
+
1013
+
1014
+
1015
+
1016
+
1017
+
1018
+
1019
+
1020
+
1021
+
1022
+
1023
+
1024
+
1025
+
1026
+
1027
+
1028
+
1029
+
1030
+
1031
+
1032
+
1033
+
1034
+
1035
+
1036
+
1037
+
1038
+
1039
+
1040
+
1041
+
1042
+
1043
+
1044
+
1045
+
1046
+
1047
+
1048
+
1049
+
1050
+
1051
+
1052
+
1053
+
1054
+
1055
+
1056
+
1057
+
1058
+
1059
+
1060
+
1061
+
1062
+
1063
+
1064
+
1065
+
1066
+
1067
+
1068
+
1069
+
1070
+
1071
+
1072
+
1073
+
1074
+
1075
+
1076
+
1077
+
1078
+
1079
+
1080
+
1081
+
1082
+
1083
+
1084
+
1085
+
1086
+
1087
+
1088
+
1089
+
1090
+
1091
+
1092
+
1093
+
1094
+
1095
+
1096
+
1097
+
1098
+
1099
+
1100
+
1101
+
1102
+
1103
+
1104
+
1105
+
1106
+
1107
+
1108
+
1109
+
1110
+
1111
+
1112
+
1113
+
1114
+
1115
+
1116
+
1117
+
1118
+
1119
+
1120
+
1121
+
1122
+
1123
+
1124
+
1125
+
1126
+
1127
+
1128
+
1129
+
1130
+
1131
+
1132
+
1133
+
1134
+
1135
+
1136
+
1137
+
1138
+
1139
+
1140
+
1141
+
1142
+
1143
+
1144
+
1145
+
1146
+
1147
+
1148
+
1149
+
1150
+
1151
+
1152
+
1153
+
1154
+
1155
+
1156
+
1157
+
1158
+
1159
+
1160
+
1161
+
1162
+
1163
+
1164
+
1165
+
1166
+
1167
+
1168
+
1169
+
1170
+
1171
+
1172
+
1173
+
1174
+
1175
+
1176
+
1177
+
1178
+
1179
+
1180
+
1181
+
1182
+
1183
+
1184
+
1185
+
1186
+
1187
+
1188
+
1189
+
1190
+
1191
+
1192
+
1193
+
1194
+
1195
+
1196
+
1197
+
1198
+
1199
+
1200
+
1201
+
1202
+
1203
+
1204
+
1205
+
1206
+
1207
+
1208
+
1209
+
1210
+
1211
+
1212
+
1213
+
1214
+
1215
+
1216
+
1217
+
1218
+
1219
+
1220
+
1221
+
1222
+
1223
+
1224
+
1225
+
1226
+
1227
+
1228
+
1229
+
1230
+
1231
+
1232
+
1233
+
1234
+
1235
+
1236
+
1237
+
1238
+
1239
+
1240
+
1241
+
1242
+
1243
+
1244
+
1245
+
1246
+
1247
+
1248
+
1249
+
1250
+
1251
+
1252
+
1253
+
1254
+
1255
+
1256
+
1257
+
1258
+
1259
+
1260
+
1261
+
1262
+
1263
+
1264
+
1265
+
1266
+
1267
+
1268
+
1269
+
1270
+
1271
+
1272
+
1273
+
1274
+
1275
+
1276
+
1277
+
1278
+
1279
+
1280
+
1281
+
1282
+
1283
+
1284
+
1285
+
1286
+
1287
+
1288
+
1289
+
1290
+
1291
+
1292
+
1293
+
1294
+
1295
+
1296
+
1297
+
1298
+
1299
+
1300
+
1301
+
1302
+
1303
+
1304
+
1305
+
1306
+
1307
+
1308
+
1309
+
1310
+
1311
+
1312
+
1313
+
1314
+
1315
+
1316
+
1317
+
1318
+
1319
+
1320
+
1321
+
1322
+
1323
+
1324
+
1325
+
1326
+
1327
+
1328
+
1329
+
1330
+
1331
+
1332
+
1333
+
1334
+
1335
+
1336
+
1337
+
1338
+
1339
+
1340
+
1341
+
1342
+
1343
+
1344
+
1345
+
1346
+
1347
+
1348
+
1349
+
1350
+
1351
+
1352
+
1353
+
1354
+
1355
+
1356
+
1357
+
1358
+
1359
+
1360
+
1361
+
1362
+
1363
+
1364
+
1365
+
1366
+
1367
+
1368
+
1369
+
1370
+
1371
+
1372
+
1373
+
1374
+
1375
+
1376
+
1377
+
1378
+
1379
+
1380
+
1381
+
1382
+
1383
+
1384
+
1385
+
1386
+
1387
+
1388
+
1389
+
1390
+
1391
+
1392
+
1393
+
1394
+
1395
+
1396
+
1397
+
1398
+
1399
+
1400
+
1401
+
1402
+
1403
+
1404
+
1405
+
1406
+
1407
+
1408
+
1409
+
1410
+
1411
+
1412
+
1413
+
1414
+
1415
+
1416
+
1417
+
1418
+
1419
+
1420
+
1421
+
1422
+
1423
+
1424
+
1425
+
1426
+
1427
+
1428
+
1429
+
1430
+
1431
+
1432
+
1433
+
1434
+
1435
+
1436
+
1437
+
1438
+
1439
+
1440
+
1441
+
1442
+
1443
+
1444
+
1445
+
1446
+
1447
+
1448
+
1449
+
1450
+
1451
+
1452
+
1453
+
1454
+
1455
+
1456
+
1457
+
1458
+
1459
+
1460
+
1461
+
1462
+
1463
+
1464
+
1465
+
1466
+
1467
+
1468
+
1469
+
1470
+
1471
+
1472
+
1473
+
1474
+
1475
+
1476
+
1477
+
1478
+
1479
+
1480
+
1481
+
1482
+
1483
+
1484
+
1485
+
1486
+
1487
+
1488
+
1489
+
1490
+
1491
+
1492
+
1493
+
1494
+
1495
+
1496
+
1497
+
1498
+
1499
+
1500
+
1501
+
1502
+
1503
+
1504
+
1505
+
1506
+
1507
+
1508
+
1509
+
1510
+
1511
+
1512
+
1513
+
1514
+
1515
+
1516
+
1517
+
1518
+
1519
+
1520
+
1521
+
1522
+
1523
+
1524
+
1525
+
1526
+
1527
+
1528
+
1529
+
1530
+
1531
+
1532
+
1533
+
1534
+
1535
+
1536
+
1537
+
1538
+
1539
+
1540
+
1541
+
1542
+
1543
+
1544
+
1545
+
1546
+
1547
+
1548
+
1549
+
1550
+
1551
+
1552
+
1553
+
1554
+
1555
+
1556
+
1557
+
1558
+
1559
+
1560
+
1561
+
1562
+
1563
+
1564
+
1565
+
1566
+
1567
+
1568
+
1569
+
1570
+
1571
+
1572
+
1573
+
1574
+
1575
+
1576
+
1577
+
1578
+
1579
+
1580
+
1581
+
1582
+
1583
+
1584
+
1585
+
1586
+
1587
+
1588
+
1589
+
1590
+
1591
+
1592
+
1593
+
1594
+
1595
+
1596
+
1597
+
1598
+
1599
+
1600
+
1601
+
1602
+
1603
+
1604
+
1605
+
1606
+
1607
+
1608
+
1609
+
1610
+
1611
+
1612
+
1613
+
1614
+
1615
+
1616
+
1617
+
1618
+
1619
+
1620
+
1621
+
1622
+
1623
+
1624
+
1625
+
1626
+
1627
+
1628
+
1629
+
1630
+
1631
+
1632
+
1633
+
1634
+
1635
+
1636
+
1637
+
1638
+
1639
+
1640
+
1641
+
1642
+
1643
+
1644
+
1645
+
1646
+
1647
+
1648
+
1649
+
1650
+
1651
+
1652
+
1653
+
1654
+
1655
+
1656
+
1657
+
1658
+
1659
+
1660
+
1661
+
1662
+
1663
+
1664
+
1665
+
1666
+
1667
+
1668
+
1669
+
1670
+
1671
+
1672
+
1673
+
1674
+
1675
+
1676
+
1677
+
1678
+
1679
+
1680
+
1681
+
1682
+
1683
+
1684
+
1685
+
1686
+
1687
+
1688
+
1689
+
1690
+
1691
+
1692
+
1693
+
1694
+
1695
+
1696
+
1697
+
1698
+
1699
+
1700
+
1701
+
1702
+
1703
+
1704
+
1705
+
1706
+
1707
+
1708
+
1709
+
1710
+
1711
+
1712
+
1713
+
1714
+
1715
+
1716
+
1717
+
1718
+ File "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py", line 819, in <module>█████████████████████████████████████████▏ | 3297/3420 [43:23<01:41, 1.21it/s]
1719
+ main()
1720
+ File "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py", line 770, in main
1721
+ train_result = trainer.train(resume_from_checkpoint=checkpoint)
1722
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 1497, in train
1723
+ self._maybe_log_save_evaluate(tr_loss, model, trial, epoch, ignore_keys_for_eval)
1724
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 1624, in _maybe_log_save_evaluate
1725
+ metrics = self.evaluate(ignore_keys=ignore_keys_for_eval)
1726
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 2284, in evaluate
1727
+ output = eval_loop(
1728
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/trainer.py", line 2448, in evaluation_loop
1729
+ for step, inputs in enumerate(dataloader):
1730
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 530, in __next__
1731
+ data = self._next_data()
1732
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 570, in _next_data
1733
+ data = self._dataset_fetcher.fetch(index) # may raise StopIteration
1734
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py", line 52, in fetch
1735
+ return self.collate_fn(data)
1736
+ File "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py", line 281, in __call__
1737
+ batch = self.processor.pad(
1738
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/models/wav2vec2/processing_wav2vec2.py", line 82, in pad
1739
+ return self.current_processor.pad(*args, **kwargs)
1740
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/feature_extraction_sequence_utils.py", line 178, in pad
1741
+ processed_features[key] = [to_numpy(v) for v in value]
1742
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/feature_extraction_sequence_utils.py", line 178, in <listcomp>
1743
+ processed_features[key] = [to_numpy(v) for v in value]
1744
+ File "/mnt/lv_ai_1_dante/ml/rolvb/venv/lib/python3.9/site-packages/transformers/utils/generic.py", line 135, in to_numpy
1745
+ return np.array(obj)
1746
+ KeyboardInterrupt
wandb/run-20220523_103002-wygrs7tw/files/requirements.txt ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ aiohttp==3.8.1
2
+ aiosignal==1.2.0
3
+ appdirs==1.4.4
4
+ async-timeout==4.0.2
5
+ attrs==21.4.0
6
+ audioread==2.1.9
7
+ certifi==2021.10.8
8
+ cffi==1.15.0
9
+ charset-normalizer==2.0.12
10
+ click==8.1.2
11
+ datasets==2.1.0
12
+ decorator==5.1.1
13
+ dill==0.3.4
14
+ docker-pycreds==0.4.0
15
+ filelock==3.6.0
16
+ frozenlist==1.3.0
17
+ fsspec==2022.3.0
18
+ gitdb==4.0.9
19
+ gitpython==3.1.27
20
+ huggingface-hub==0.5.1
21
+ hypothesis==6.46.5
22
+ idna==3.3
23
+ jiwer==2.3.0
24
+ joblib==1.1.0
25
+ kenlm==0.0.0
26
+ librosa==0.9.1
27
+ llvmlite==0.38.0
28
+ multidict==6.0.2
29
+ multiprocess==0.70.12.2
30
+ numba==0.55.1
31
+ numpy==1.21.6
32
+ packaging==21.3
33
+ pandas==1.4.2
34
+ pathtools==0.1.2
35
+ pillow==9.1.0
36
+ pip==20.3.4
37
+ pkg-resources==0.0.0
38
+ pooch==1.6.0
39
+ promise==2.3
40
+ protobuf==3.20.1
41
+ psutil==5.9.0
42
+ pyarrow==7.0.0
43
+ pycparser==2.21
44
+ pyctcdecode==0.3.0
45
+ pygtrie==2.4.2
46
+ pyparsing==3.0.8
47
+ python-dateutil==2.8.2
48
+ python-levenshtein==0.12.2
49
+ pytz==2022.1
50
+ pyyaml==6.0
51
+ regex==2022.4.24
52
+ requests==2.27.1
53
+ resampy==0.2.2
54
+ responses==0.18.0
55
+ sacremoses==0.0.49
56
+ scikit-learn==1.0.2
57
+ scipy==1.8.0
58
+ sentry-sdk==1.5.10
59
+ setproctitle==1.2.3
60
+ setuptools==44.1.1
61
+ shortuuid==1.0.8
62
+ six==1.16.0
63
+ smmap==5.0.0
64
+ sortedcontainers==2.4.0
65
+ soundfile==0.10.3.post1
66
+ threadpoolctl==3.1.0
67
+ tokenizers==0.12.1
68
+ torch==1.11.0+cu113
69
+ torchaudio==0.11.0+cu113
70
+ torchvision==0.12.0+cu113
71
+ tqdm==4.64.0
72
+ transformers==4.18.0
73
+ typing-extensions==4.2.0
74
+ urllib3==1.26.9
75
+ wandb==0.12.15
76
+ xxhash==3.0.0
77
+ yarl==1.7.2
wandb/run-20220523_103002-wygrs7tw/files/wandb-metadata.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "os": "Linux-5.13.0-40-generic-x86_64-with-glibc2.34",
3
+ "python": "3.9.7",
4
+ "heartbeatAt": "2022-05-23T08:30:03.662767",
5
+ "startedAt": "2022-05-23T08:30:02.486763",
6
+ "docker": null,
7
+ "cpu_count": 96,
8
+ "cuda": null,
9
+ "args": [
10
+ "--model_name_or_path=facebook/wav2vec2-xls-r-1b",
11
+ "--hub_model_id=NbAiLab/wav2vec2-1b-npsc-nst-bokmaal",
12
+ "--output_dir=./",
13
+ "--overwrite_output_dir",
14
+ "--num_train_epochs=40",
15
+ "--per_device_train_batch_size=12",
16
+ "--per_device_eval_batch_size=12",
17
+ "--gradient_accumulation_steps=2",
18
+ "--learning_rate=2e-5",
19
+ "--warmup_steps=2000",
20
+ "--length_column_name=input_length",
21
+ "--evaluation_strategy=steps",
22
+ "--text_column_name=text",
23
+ "--save_steps=500",
24
+ "--eval_steps=500",
25
+ "--logging_steps=100",
26
+ "--layerdrop=0.041",
27
+ "--attention_dropout=0.094",
28
+ "--activation_dropout=0.055",
29
+ "--hidden_dropout=0.047",
30
+ "--save_total_limit=3",
31
+ "--freeze_feature_encoder",
32
+ "--feat_proj_dropout=0.04",
33
+ "--mask_time_prob=0.082",
34
+ "--mask_time_length=10",
35
+ "--mask_feature_prob=0.25",
36
+ "--mask_feature_length=64",
37
+ "--gradient_checkpointing",
38
+ "--min_duration_in_seconds=0.5",
39
+ "--max_duration_in_seconds=30.0",
40
+ "--use_auth_token",
41
+ "--seed=42",
42
+ "--fp16",
43
+ "--group_by_length",
44
+ "--do_train",
45
+ "--do_eval",
46
+ "--push_to_hub",
47
+ "--preprocessing_num_workers=32",
48
+ "--ctc_zero_infinity"
49
+ ],
50
+ "state": "running",
51
+ "program": "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py",
52
+ "codePath": "run_speech_recognition_ctc.py",
53
+ "git": {
54
+ "remote": "https://huggingface.co/NbAiLab/wav2vec2-1b-npsc-nst-bokmaal",
55
+ "commit": "05ba5dd22c9cec7365fba75fae8630eec1302aff"
56
+ },
57
+ "email": "[email protected]",
58
+ "root": "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal",
59
+ "host": "dante",
60
+ "username": "rolvb",
61
+ "executable": "/mnt/lv_ai_1_dante/ml/rolvb/venv/bin/python"
62
+ }
wandb/run-20220523_103002-wygrs7tw/files/wandb-summary.json ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_103002-wygrs7tw/logs/debug-internal.log ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_103002-wygrs7tw/logs/debug.log ADDED
@@ -0,0 +1,181 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_setup.py:_flush():75] Loading settings from /home/rolvb/.config/wandb/settings
2
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_setup.py:_flush():75] Loading settings from /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/settings
3
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_setup.py:_flush():75] Loading settings from environment variables: {'project': 'wav2vec2', 'entity': 'NbAiLab'}
4
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_setup.py:_flush():75] Inferring run settings from compute environment: {'program_relpath': 'run_speech_recognition_ctc.py', 'program': '/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py'}
5
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_init.py:_log_setup():437] Logging user logs to /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/run-20220523_103002-wygrs7tw/logs/debug.log
6
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_init.py:_log_setup():438] Logging internal logs to /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/run-20220523_103002-wygrs7tw/logs/debug-internal.log
7
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_init.py:init():471] calling init triggers
8
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_init.py:init():474] wandb.init called with sweep_config: {}
9
+ config: {}
10
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [wandb_init.py:init():524] starting backend
11
+ 2022-05-23 10:30:02,503 INFO MainThread:1047787 [backend.py:_multiprocessing_setup():97] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
12
+ 2022-05-23 10:30:02,593 INFO MainThread:1047787 [backend.py:ensure_launched():217] starting backend process...
13
+ 2022-05-23 10:30:02,734 INFO MainThread:1047787 [backend.py:ensure_launched():222] started backend process with pid: 1048602
14
+ 2022-05-23 10:30:02,736 INFO MainThread:1047787 [wandb_init.py:init():533] backend started and connected
15
+ 2022-05-23 10:30:02,745 INFO MainThread:1047787 [wandb_init.py:init():597] updated telemetry
16
+ 2022-05-23 10:30:03,009 INFO MainThread:1047787 [wandb_init.py:init():628] communicating run to backend with 30 second timeout
17
+ 2022-05-23 10:30:03,518 INFO MainThread:1047787 [wandb_run.py:_on_init():1923] communicating current version
18
+ 2022-05-23 10:30:03,647 INFO MainThread:1047787 [wandb_run.py:_on_init():1927] got version response upgrade_message: "wandb version 0.12.16 is available! To upgrade, please run:\n $ pip install wandb --upgrade"
19
+
20
+ 2022-05-23 10:30:03,648 INFO MainThread:1047787 [wandb_init.py:init():659] starting run threads in backend
21
+ 2022-05-23 10:30:03,698 INFO MainThread:1047787 [wandb_run.py:_console_start():1897] atexit reg
22
+ 2022-05-23 10:30:03,699 INFO MainThread:1047787 [wandb_run.py:_redirect():1770] redirect: SettingsConsole.REDIRECT
23
+ 2022-05-23 10:30:03,699 INFO MainThread:1047787 [wandb_run.py:_redirect():1775] Redirecting console.
24
+ 2022-05-23 10:30:03,702 INFO MainThread:1047787 [wandb_run.py:_redirect():1831] Redirects installed.
25
+ 2022-05-23 10:30:03,702 INFO MainThread:1047787 [wandb_init.py:init():684] run started, returning control to user process
26
+ 2022-05-23 10:30:03,725 INFO MainThread:1047787 [wandb_run.py:_config_callback():1131] config_cb None None {'return_dict': True, 'output_hidden_states': False, 'output_attentions': False, 'torchscript': False, 'torch_dtype': 'float32', 'use_bfloat16': False, 'pruned_heads': {}, 'tie_word_embeddings': True, 'is_encoder_decoder': False, 'is_decoder': False, 'cross_attention_hidden_size': None, 'add_cross_attention': False, 'tie_encoder_decoder': False, 'max_length': 20, 'min_length': 0, 'do_sample': False, 'early_stopping': False, 'num_beams': 1, 'num_beam_groups': 1, 'diversity_penalty': 0.0, 'temperature': 1.0, 'top_k': 50, 'top_p': 1.0, 'typical_p': 1.0, 'repetition_penalty': 1.0, 'length_penalty': 1.0, 'no_repeat_ngram_size': 0, 'encoder_no_repeat_ngram_size': 0, 'bad_words_ids': None, 'num_return_sequences': 1, 'chunk_size_feed_forward': 0, 'output_scores': False, 'return_dict_in_generate': False, 'forced_bos_token_id': None, 'forced_eos_token_id': None, 'remove_invalid_values': False, 'exponential_decay_length_penalty': None, 'architectures': ['Wav2Vec2ForPreTraining'], 'finetuning_task': None, 'id2label': {0: 'LABEL_0', 1: 'LABEL_1'}, 'label2id': {'LABEL_0': 0, 'LABEL_1': 1}, 'tokenizer_class': None, 'prefix': None, 'bos_token_id': 1, 'pad_token_id': 31, 'eos_token_id': 2, 'sep_token_id': None, 'decoder_start_token_id': None, 'task_specific_params': None, 'problem_type': None, '_name_or_path': 'facebook/wav2vec2-xls-r-1b', 'transformers_version': '4.18.0', 'feat_extract_dropout': 0.0, 'model_type': 'wav2vec2', 'num_feat_extract_layers': 7, 'hidden_size': 1280, 'feat_extract_norm': 'layer', 'feat_extract_activation': 'gelu', 'conv_dim': [512, 512, 512, 512, 512, 512, 512], 'conv_stride': [5, 2, 2, 2, 2, 2, 2], 'conv_kernel': [10, 3, 3, 3, 3, 2, 2], 'conv_bias': True, 'num_conv_pos_embeddings': 128, 'num_conv_pos_embedding_groups': 16, 'num_hidden_layers': 48, 'intermediate_size': 5120, 'hidden_act': 'gelu', 'num_attention_heads': 16, 'hidden_dropout': 0.047, 'attention_dropout': 0.094, 'activation_dropout': 0.055, 'feat_proj_dropout': 0.04, 'final_dropout': 0.0, 'layerdrop': 0.041, 'layer_norm_eps': 1e-05, 'initializer_range': 0.02, 'vocab_size': 34, 'do_stable_layer_norm': True, 'use_weighted_layer_sum': False, 'apply_spec_augment': True, 'mask_time_prob': 0.082, 'mask_time_length': 10, 'mask_time_min_masks': 2, 'mask_feature_prob': 0.25, 'mask_feature_length': 64, 'mask_feature_min_masks': 0, 'num_codevectors_per_group': 320, 'num_codevector_groups': 2, 'contrastive_logits_temperature': 0.1, 'feat_quantizer_dropout': 0.0, 'num_negatives': 100, 'codevector_dim': 1024, 'proj_codevector_dim': 1024, 'diversity_loss_weight': 0.1, 'ctc_loss_reduction': 'mean', 'ctc_zero_infinity': True, 'add_adapter': False, 'adapter_kernel_size': 3, 'adapter_stride': 2, 'num_adapter_layers': 3, 'output_hidden_size': 1280, 'classifier_proj_size': 256, 'tdnn_dim': [512, 512, 512, 512, 1500], 'tdnn_kernel': [5, 3, 3, 1, 1], 'tdnn_dilation': [1, 2, 3, 1, 1], 'xvector_output_dim': 512, 'output_dir': './', 'overwrite_output_dir': True, 'do_train': True, 'do_eval': True, 'do_predict': False, 'evaluation_strategy': 'steps', 'prediction_loss_only': False, 'per_device_train_batch_size': 12, 'per_device_eval_batch_size': 12, 'per_gpu_train_batch_size': 'None', 'per_gpu_eval_batch_size': 'None', 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': 'None', 'eval_delay': 0, 'learning_rate': 2e-05, 'weight_decay': 0.0, 'adam_beta1': 0.9, 'adam_beta2': 0.999, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 40.0, 'max_steps': -1, 'lr_scheduler_type': 'linear', 'warmup_ratio': 0.0, 'warmup_steps': 2000, 'log_level': -1, 'log_level_replica': -1, 'log_on_each_node': True, 'logging_dir': './runs/May23_10-06-11_dante', 'logging_strategy': 'steps', 'logging_first_step': False, 'logging_steps': 100, 'logging_nan_inf_filter': True, 'save_strategy': 'steps', 'save_steps': 500, 'save_total_limit': 3, 'save_on_each_node': False, 'no_cuda': False, 'seed': 42, 'data_seed': 'None', 'bf16': False, 'fp16': True, 'fp16_opt_level': 'O1', 'half_precision_backend': 'amp', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': 'None', 'local_rank': -1, 'xpu_backend': 'None', 'tpu_num_cores': 'None', 'tpu_metrics_debug': False, 'debug': '[]', 'dataloader_drop_last': False, 'eval_steps': 500, 'dataloader_num_workers': 0, 'past_index': -1, 'run_name': './', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': 'None', 'load_best_model_at_end': False, 'metric_for_best_model': 'None', 'greater_is_better': 'None', 'ignore_data_skip': False, 'sharded_ddp': '[]', 'deepspeed': 'None', 'label_smoothing_factor': 0.0, 'optim': 'adamw_hf', 'adafactor': False, 'group_by_length': True, 'length_column_name': 'input_length', 'report_to': "['wandb']", 'ddp_find_unused_parameters': 'None', 'ddp_bucket_cap_mb': 'None', 'dataloader_pin_memory': True, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': True, 'resume_from_checkpoint': 'None', 'hub_model_id': 'NbAiLab/wav2vec2-1b-npsc-nst-bokmaal', 'hub_strategy': 'every_save', 'hub_token': '<HUB_TOKEN>', 'gradient_checkpointing': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': 'None', 'push_to_hub_organization': 'None', 'push_to_hub_token': '<PUSH_TO_HUB_TOKEN>', '_n_gpu': 1, 'mp_parameters': '', 'train_batch_size': 12, 'eval_batch_size': 12}
27
+ 2022-05-23 10:30:03,728 INFO MainThread:1047787 [wandb_watch.py:watch():47] Watching
28
+ 2022-05-23 11:33:10,501 INFO MainThread:1047787 [wandb_run.py:_atexit_cleanup():1866] got exitcode: 255
29
+ 2022-05-23 11:33:10,503 INFO MainThread:1047787 [wandb_run.py:_restore():1838] restore
30
+ 2022-05-23 11:33:12,930 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
31
+ wandb_count: 1
32
+ }
33
+ pusher_stats {
34
+ uploaded_bytes: 2176
35
+ total_bytes: 2176
36
+ }
37
+
38
+ 2022-05-23 11:33:13,179 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
39
+ wandb_count: 1
40
+ }
41
+ pusher_stats {
42
+ uploaded_bytes: 2176
43
+ total_bytes: 2176
44
+ }
45
+
46
+ 2022-05-23 11:33:15,044 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
47
+ wandb_count: 1
48
+ }
49
+ pusher_stats {
50
+ uploaded_bytes: 2176
51
+ total_bytes: 2176
52
+ }
53
+
54
+ 2022-05-23 11:33:15,184 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
55
+ wandb_count: 5
56
+ }
57
+ pusher_stats {
58
+ uploaded_bytes: 2176
59
+ total_bytes: 1778293
60
+ }
61
+
62
+ 2022-05-23 11:33:15,286 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
63
+ wandb_count: 5
64
+ }
65
+ pusher_stats {
66
+ uploaded_bytes: 2176
67
+ total_bytes: 1778293
68
+ }
69
+
70
+ 2022-05-23 11:33:15,388 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
71
+ wandb_count: 5
72
+ }
73
+ pusher_stats {
74
+ uploaded_bytes: 2176
75
+ total_bytes: 1778293
76
+ }
77
+
78
+ 2022-05-23 11:33:15,490 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
79
+ wandb_count: 5
80
+ }
81
+ pusher_stats {
82
+ uploaded_bytes: 2176
83
+ total_bytes: 1778293
84
+ }
85
+
86
+ 2022-05-23 11:33:15,593 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
87
+ wandb_count: 5
88
+ }
89
+ pusher_stats {
90
+ uploaded_bytes: 561382
91
+ total_bytes: 1778293
92
+ }
93
+
94
+ 2022-05-23 11:33:15,695 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
95
+ wandb_count: 5
96
+ }
97
+ pusher_stats {
98
+ uploaded_bytes: 1778293
99
+ total_bytes: 1778293
100
+ }
101
+
102
+ 2022-05-23 11:33:15,797 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
103
+ wandb_count: 5
104
+ }
105
+ pusher_stats {
106
+ uploaded_bytes: 1778293
107
+ total_bytes: 1778293
108
+ }
109
+
110
+ 2022-05-23 11:33:15,899 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
111
+ wandb_count: 5
112
+ }
113
+ pusher_stats {
114
+ uploaded_bytes: 1778293
115
+ total_bytes: 1778293
116
+ }
117
+
118
+ 2022-05-23 11:33:16,001 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
119
+ wandb_count: 5
120
+ }
121
+ pusher_stats {
122
+ uploaded_bytes: 1778293
123
+ total_bytes: 1778293
124
+ }
125
+
126
+ 2022-05-23 11:33:16,103 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
127
+ wandb_count: 5
128
+ }
129
+ pusher_stats {
130
+ uploaded_bytes: 1778293
131
+ total_bytes: 1778293
132
+ }
133
+
134
+ 2022-05-23 11:33:16,206 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
135
+ wandb_count: 5
136
+ }
137
+ pusher_stats {
138
+ uploaded_bytes: 1778293
139
+ total_bytes: 1778293
140
+ }
141
+
142
+ 2022-05-23 11:33:16,308 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
143
+ wandb_count: 5
144
+ }
145
+ pusher_stats {
146
+ uploaded_bytes: 1778293
147
+ total_bytes: 1778293
148
+ }
149
+
150
+ 2022-05-23 11:33:16,410 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
151
+ wandb_count: 5
152
+ }
153
+ pusher_stats {
154
+ uploaded_bytes: 1778293
155
+ total_bytes: 1778293
156
+ }
157
+
158
+ 2022-05-23 11:33:18,038 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: file_counts {
159
+ wandb_count: 5
160
+ }
161
+ pusher_stats {
162
+ uploaded_bytes: 1778293
163
+ total_bytes: 1778293
164
+ }
165
+
166
+ 2022-05-23 11:33:18,596 INFO MainThread:1047787 [wandb_run.py:_on_finish():1995] got exit ret: done: true
167
+ exit_result {
168
+ }
169
+ file_counts {
170
+ wandb_count: 5
171
+ }
172
+ pusher_stats {
173
+ uploaded_bytes: 1778293
174
+ total_bytes: 1778293
175
+ }
176
+ local_info {
177
+ }
178
+
179
+ 2022-05-23 11:33:19,771 INFO MainThread:1047787 [wandb_run.py:_footer_history_summary_info():3102] rendering history
180
+ 2022-05-23 11:33:19,820 INFO MainThread:1047787 [wandb_run.py:_footer_history_summary_info():3134] rendering summary
181
+ 2022-05-23 11:33:19,824 INFO MainThread:1047787 [wandb_run.py:_footer_sync_info():3057] logging synced files
wandb/run-20220523_103002-wygrs7tw/run-wygrs7tw.wandb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e3b04296f0f8a6544a2af266cc3f7b5ced5fdd8010b5089ff3bd1dc0848f487
3
+ size 9407928
wandb/run-20220523_115145-3dybzmyz/files/config.yaml ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_115145-3dybzmyz/files/output.log ADDED
@@ -0,0 +1,1788 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+
3
+
4
+
5
+
6
+
7
+
8
+
9
+
10
+
11
+
12
+
13
+
14
+
15
+
16
+
17
+
18
+
19
+
20
+
21
+
22
+
23
+
24
+
25
+
26
+
27
+
28
+
29
+
30
+
31
+
32
+
33
+
34
+
35
+
36
+
37
+
38
+
39
+
40
+
41
+
42
+
43
+
44
+
45
+
46
+
47
+
48
+
49
+
50
+
51
+
52
+
53
+
54
+
55
+
56
+
57
+
58
+
59
+
60
+
61
+
62
+
63
+
64
+
65
+
66
+
67
+
68
+
69
+
70
+
71
+
72
+
73
+
74
+
75
+
76
+
77
+
78
+
79
+
80
+
81
+
82
+
83
+
84
+
85
+
86
+ 0%| | 99/499680 [03:57<142:58:51, 1.03s/it]
87
+
88
+
89
+
90
+
91
+
92
+
93
+
94
+
95
+
96
+
97
+
98
+
99
+
100
+
101
+
102
+
103
+
104
+
105
+
106
+
107
+
108
+
109
+
110
+
111
+
112
+
113
+
114
+
115
+
116
+
117
+
118
+
119
+
120
+
121
+
122
+
123
+
124
+
125
+
126
+
127
+
128
+
129
+
130
+
131
+
132
+
133
+
134
+
135
+
136
+
137
+
138
+
139
+
140
+
141
+
142
+
143
+
144
+
145
+
146
+
147
+
148
+
149
+
150
+
151
+
152
+
153
+
154
+
155
+
156
+
157
+
158
+
159
+
160
+
161
+
162
+
163
+
164
+
165
+
166
+
167
+
168
+
169
+
170
+
171
+
172
+
173
+
174
+ 0%| | 200/499680 [07:55<131:47:55, 1.05it/s]
175
+
176
+
177
+
178
+
179
+
180
+
181
+
182
+
183
+
184
+
185
+
186
+
187
+
188
+
189
+
190
+
191
+
192
+
193
+
194
+
195
+
196
+
197
+
198
+
199
+
200
+
201
+
202
+
203
+
204
+
205
+
206
+
207
+
208
+
209
+
210
+
211
+
212
+
213
+
214
+
215
+
216
+
217
+
218
+
219
+
220
+
221
+
222
+
223
+
224
+
225
+
226
+
227
+
228
+
229
+
230
+
231
+
232
+
233
+
234
+
235
+
236
+
237
+
238
+
239
+
240
+
241
+
242
+
243
+
244
+
245
+
246
+
247
+
248
+
249
+
250
+
251
+
252
+
253
+
254
+
255
+
256
+
257
+
258
+
259
+
260
+ 0%| | 298/499680 [11:52<147:05:38, 1.06s/it]
261
+
262
+
263
+
264
+
265
+
266
+
267
+
268
+
269
+
270
+
271
+
272
+
273
+
274
+
275
+
276
+
277
+
278
+
279
+
280
+
281
+
282
+
283
+
284
+
285
+
286
+
287
+
288
+
289
+
290
+
291
+
292
+
293
+
294
+
295
+
296
+
297
+
298
+
299
+
300
+
301
+
302
+
303
+
304
+
305
+
306
+
307
+
308
+
309
+
310
+
311
+
312
+
313
+
314
+
315
+
316
+
317
+
318
+
319
+
320
+
321
+
322
+
323
+
324
+
325
+
326
+
327
+
328
+
329
+
330
+
331
+
332
+
333
+
334
+
335
+
336
+
337
+
338
+
339
+
340
+
341
+
342
+
343
+
344
+
345
+
346
+
347
+ 0%| | 399/499680 [15:57<146:33:14, 1.06s/it]
348
+
349
+
350
+
351
+
352
+
353
+
354
+
355
+
356
+
357
+
358
+
359
+
360
+
361
+
362
+
363
+
364
+
365
+
366
+
367
+
368
+
369
+
370
+
371
+
372
+
373
+
374
+
375
+
376
+
377
+
378
+
379
+
380
+
381
+
382
+
383
+
384
+
385
+
386
+
387
+
388
+
389
+
390
+
391
+
392
+
393
+
394
+
395
+
396
+
397
+
398
+
399
+
400
+
401
+
402
+
403
+
404
+
405
+
406
+
407
+
408
+
409
+
410
+
411
+
412
+
413
+
414
+
415
+
416
+
417
+
418
+
419
+
420
+
421
+
422
+
423
+
424
+
425
+
426
+
427
+
428
+
429
+
430
+
431
+
432
+
433
+
434
+
435
+ 0%| | 498/499680 [20:03<149:47:46, 1.08s/it]
436
+ 0%| | 500/499680 [20:05<135:25:31, 1.02it/s]The following columns in the evaluation set don't have a corresponding argument in `Wav2Vec2ForCTC.forward` and have been ignored: input_length. If input_length are not expected by `Wav2Vec2ForCTC.forward`, you can safely ignore this message.
437
+ ***** Running Evaluation *****
438
+ Num examples = 40740
439
+ Batch size = 12
440
+
441
+
442
+
443
+
444
+
445
+
446
+
447
+
448
+
449
+
450
+
451
+
452
+
453
+
454
+
455
+
456
+
457
+
458
+
459
+
460
+
461
+
462
+
463
+
464
+
465
+
466
+
467
+
468
+
469
+
470
+
471
+
472
+
473
+
474
+
475
+
476
+
477
+
478
+
479
+
480
+
481
+
482
+
483
+
484
+
485
+
486
+
487
+
488
+
489
+
490
+
491
+
492
+
493
+
494
+
495
+
496
+
497
+
498
+
499
+
500
+
501
+
502
+
503
+
504
+
505
+
506
+
507
+
508
+
509
+
510
+
511
+
512
+
513
+
514
+
515
+
516
+
517
+
518
+
519
+
520
+
521
+
522
+
523
+
524
+
525
+
526
+
527
+
528
+
529
+
530
+
531
+
532
+
533
+
534
+
535
+
536
+
537
+
538
+
539
+
540
+
541
+
542
+
543
+
544
+
545
+
546
+
547
+
548
+
549
+
550
+
551
+
552
+
553
+
554
+
555
+
556
+
557
+
558
+
559
+
560
+
561
+
562
+
563
+
564
+
565
+
566
+
567
+
568
+
569
+
570
+
571
+
572
+
573
+
574
+
575
+
576
+
577
+
578
+
579
+
580
+
581
+
582
+
583
+
584
+
585
+
586
+
587
+
588
+
589
+
590
+
591
+
592
+
593
+
594
+
595
+
596
+
597
+
598
+
599
+
600
+
601
+
602
+
603
+
604
+
605
+
606
+
607
+
608
+
609
+
610
+
611
+
612
+
613
+
614
+
615
+
616
+
617
+
618
+
619
+
620
+
621
+
622
+
623
+
624
+
625
+
626
+
627
+
628
+
629
+
630
+
631
+
632
+
633
+
634
+
635
+
636
+
637
+
638
+
639
+
640
+
641
+
642
+
643
+
644
+
645
+
646
+
647
+
648
+
649
+
650
+
651
+
652
+
653
+
654
+
655
+
656
+
657
+
658
+
659
+
660
+
661
+
662
+
663
+
664
+
665
+
666
+
667
+
668
+
669
+
670
+
671
+
672
+
673
+
674
+
675
+
676
+
677
+
678
+
679
+
680
+
681
+
682
+
683
+
684
+
685
+
686
+
687
+
688
+
689
+
690
+
691
+
692
+
693
+
694
+
695
+
696
+
697
+
698
+
699
+
700
+
701
+
702
+
703
+
704
+
705
+
706
+
707
+
708
+
709
+
710
+
711
+
712
+
713
+
714
+
715
+
716
+
717
+
718
+
719
+
720
+
721
+
722
+
723
+
724
+
725
+
726
+
727
+
728
+
729
+
730
+
731
+
732
+
733
+
734
+
735
+
736
+
737
+
738
+
739
+
740
+
741
+
742
+
743
+
744
+
745
+
746
+
747
+
748
+
749
+
750
+
751
+
752
+
753
+
754
+
755
+
756
+
757
+
758
+
759
+
760
+
761
+
762
+
763
+
764
+
765
+
766
+
767
+
768
+
769
+
770
+
771
+
772
+
773
+
774
+
775
+
776
+
777
+
778
+
779
+
780
+
781
+
782
+
783
+
784
+
785
+
786
+
787
+
788
+
789
+
790
+
791
+
792
+
793
+
794
+
795
+
796
+
797
+
798
+
799
+
800
+
801
+
802
+
803
+
804
+
805
+
806
+
807
+
808
+
809
+
810
+
811
+
812
+
813
+
814
+
815
+
816
+
817
+
818
+
819
+
820
+
821
+
822
+
823
+
824
+
825
+
826
+
827
+
828
+
829
+
830
+
831
+
832
+
833
+
834
+
835
+
836
+
837
+
838
+
839
+
840
+
841
+
842
+
843
+
844
+
845
+
846
+
847
+
848
+
849
+
850
+
851
+
852
+
853
+
854
+
855
+
856
+
857
+
858
+
859
+
860
+
861
+
862
+
863
+
864
+
865
+
866
+
867
+
868
+
869
+
870
+
871
+
872
+
873
+
874
+
875
+
876
+
877
+
878
+
879
+
880
+
881
+
882
+
883
+
884
+
885
+
886
+
887
+
888
+
889
+
890
+
891
+
892
+
893
+
894
+
895
+
896
+
897
+
898
+
899
+
900
+
901
+
902
+
903
+
904
+
905
+
906
+
907
+
908
+
909
+
910
+
911
+
912
+
913
+
914
+
915
+
916
+
917
+
918
+
919
+
920
+
921
+
922
+
923
+
924
+
925
+
926
+
927
+
928
+
929
+
930
+
931
+
932
+
933
+
934
+
935
+
936
+
937
+
938
+
939
+
940
+
941
+
942
+
943
+
944
+
945
+
946
+
947
+
948
+
949
+
950
+
951
+
952
+
953
+
954
+
955
+
956
+
957
+
958
+
959
+
960
+
961
+
962
+
963
+
964
+
965
+
966
+
967
+
968
+
969
+
970
+
971
+
972
+
973
+
974
+
975
+
976
+
977
+
978
+
979
+
980
+
981
+
982
+
983
+
984
+
985
+
986
+
987
+
988
+
989
+
990
+
991
+
992
+
993
+
994
+
995
+
996
+
997
+
998
+
999
+
1000
+
1001
+
1002
+
1003
+
1004
+
1005
+
1006
+
1007
+
1008
+
1009
+
1010
+
1011
+
1012
+
1013
+
1014
+
1015
+
1016
+
1017
+
1018
+
1019
+
1020
+
1021
+
1022
+
1023
+
1024
+
1025
+
1026
+
1027
+
1028
+
1029
+
1030
+
1031
+
1032
+
1033
+
1034
+
1035
+
1036
+
1037
+
1038
+
1039
+
1040
+
1041
+
1042
+
1043
+
1044
+
1045
+
1046
+
1047
+
1048
+
1049
+
1050
+
1051
+
1052
+
1053
+
1054
+
1055
+
1056
+
1057
+
1058
+
1059
+
1060
+
1061
+
1062
+
1063
+
1064
+
1065
+
1066
+
1067
+
1068
+
1069
+
1070
+
1071
+
1072
+
1073
+
1074
+
1075
+
1076
+
1077
+
1078
+
1079
+
1080
+
1081
+
1082
+
1083
+
1084
+
1085
+
1086
+
1087
+
1088
+
1089
+
1090
+
1091
+
1092
+
1093
+
1094
+
1095
+
1096
+
1097
+
1098
+
1099
+
1100
+
1101
+
1102
+
1103
+
1104
+
1105
+
1106
+
1107
+
1108
+
1109
+
1110
+
1111
+
1112
+
1113
+
1114
+
1115
+
1116
+
1117
+
1118
+
1119
+
1120
+
1121
+
1122
+
1123
+
1124
+
1125
+
1126
+
1127
+
1128
+
1129
+
1130
+
1131
+
1132
+
1133
+
1134
+
1135
+
1136
+
1137
+
1138
+
1139
+
1140
+
1141
+
1142
+
1143
+
1144
+
1145
+
1146
+
1147
+
1148
+
1149
+
1150
+
1151
+
1152
+
1153
+
1154
+
1155
+
1156
+
1157
+
1158
+
1159
+
1160
+
1161
+
1162
+
1163
+
1164
+
1165
+
1166
+
1167
+
1168
+
1169
+
1170
+
1171
+
1172
+
1173
+
1174
+
1175
+
1176
+
1177
+
1178
+
1179
+
1180
+
1181
+
1182
+
1183
+
1184
+
1185
+
1186
+
1187
+
1188
+
1189
+
1190
+
1191
+
1192
+
1193
+
1194
+
1195
+
1196
+
1197
+
1198
+
1199
+
1200
+
1201
+
1202
+
1203
+
1204
+
1205
+
1206
+
1207
+
1208
+
1209
+
1210
+
1211
+
1212
+
1213
+
1214
+
1215
+
1216
+
1217
+
1218
+
1219
+
1220
+
1221
+
1222
+
1223
+
1224
+
1225
+
1226
+
1227
+
1228
+
1229
+
1230
+
1231
+
1232
+
1233
+
1234
+
1235
+
1236
+
1237
+
1238
+
1239
+
1240
+
1241
+
1242
+
1243
+
1244
+
1245
+
1246
+
1247
+
1248
+
1249
+
1250
+
1251
+
1252
+
1253
+
1254
+
1255
+
1256
+
1257
+
1258
+
1259
+
1260
+
1261
+
1262
+
1263
+
1264
+
1265
+
1266
+
1267
+
1268
+
1269
+
1270
+
1271
+
1272
+
1273
+
1274
+
1275
+
1276
+
1277
+
1278
+
1279
+
1280
+
1281
+
1282
+
1283
+
1284
+
1285
+
1286
+
1287
+
1288
+
1289
+
1290
+
1291
+
1292
+
1293
+
1294
+
1295
+
1296
+
1297
+
1298
+
1299
+
1300
+
1301
+
1302
+
1303
+
1304
+
1305
+
1306
+
1307
+
1308
+
1309
+
1310
+
1311
+
1312
+
1313
+
1314
+
1315
+
1316
+
1317
+
1318
+
1319
+
1320
+
1321
+
1322
+
1323
+
1324
+
1325
+
1326
+
1327
+
1328
+
1329
+
1330
+
1331
+
1332
+
1333
+
1334
+
1335
+
1336
+
1337
+
1338
+
1339
+
1340
+
1341
+
1342
+
1343
+
1344
+
1345
+
1346
+
1347
+
1348
+
1349
+
1350
+
1351
+
1352
+
1353
+
1354
+
1355
+
1356
+
1357
+
1358
+
1359
+
1360
+
1361
+
1362
+
1363
+
1364
+
1365
+
1366
+
1367
+
1368
+
1369
+
1370
+
1371
+
1372
+
1373
+
1374
+
1375
+
1376
+
1377
+
1378
+
1379
+
1380
+
1381
+
1382
+
1383
+
1384
+
1385
+
1386
+
1387
+
1388
+
1389
+
1390
+
1391
+
1392
+
1393
+
1394
+
1395
+
1396
+
1397
+
1398
+
1399
+
1400
+
1401
+
1402
+
1403
+
1404
+
1405
+
1406
+
1407
+
1408
+
1409
+
1410
+
1411
+
1412
+
1413
+
1414
+
1415
+
1416
+
1417
+
1418
+
1419
+
1420
+
1421
+
1422
+
1423
+
1424
+
1425
+
1426
+
1427
+
1428
+
1429
+
1430
+
1431
+
1432
+
1433
+
1434
+
1435
+
1436
+
1437
+
1438
+
1439
+
1440
+
1441
+
1442
+
1443
+
1444
+
1445
+
1446
+
1447
+
1448
+
1449
+
1450
+
1451
+
1452
+
1453
+
1454
+
1455
+
1456
+
1457
+
1458
+
1459
+
1460
+
1461
+
1462
+
1463
+
1464
+
1465
+
1466
+
1467
+
1468
+
1469
+
1470
+
1471
+
1472
+
1473
+
1474
+
1475
+
1476
+
1477
+
1478
+
1479
+
1480
+
1481
+
1482
+
1483
+
1484
+
1485
+
1486
+
1487
+
1488
+
1489
+
1490
+
1491
+
1492
+
1493
+
1494
+
1495
+
1496
+
1497
+
1498
+
1499
+
1500
+
1501
+
1502
+
1503
+
1504
+
1505
+
1506
+
1507
+
1508
+
1509
+
1510
+
1511
+
1512
+
1513
+
1514
+
1515
+
1516
+
1517
+
1518
+
1519
+
1520
+
1521
+
1522
+
1523
+
1524
+
1525
+
1526
+
1527
+
1528
+
1529
+
1530
+
1531
+
1532
+
1533
+
1534
+
1535
+
1536
+
1537
+
1538
+
1539
+
1540
+
1541
+
1542
+
1543
+
1544
+
1545
+
1546
+
1547
+
1548
+
1549
+
1550
+
1551
+
1552
+
1553
+
1554
+
1555
+
1556
+
1557
+
1558
+
1559
+
1560
+
1561
+
1562
+
1563
+
1564
+
1565
+
1566
+
1567
+
1568
+
1569
+
1570
+
1571
+
1572
+
1573
+
1574
+
1575
+
1576
+
1577
+
1578
+
1579
+
1580
+
1581
+
1582
+
1583
+
1584
+
1585
+
1586
+
1587
+
1588
+
1589
+
1590
+
1591
+
1592
+
1593
+
1594
+
1595
+
1596
+
1597
+
1598
+
1599
+
1600
+
1601
+
1602
+
1603
+
1604
+
1605
+
1606
+
1607
+
1608
+
1609
+
1610
+
1611
+
1612
+
1613
+
1614
+
1615
+
1616
+
1617
+
1618
+
1619
+
1620
+
1621
+
1622
+
1623
+
1624
+
1625
+
1626
+
1627
+
1628
+
1629
+
1630
+
1631
+
1632
+
1633
+
1634
+
1635
+
1636
+
1637
+
1638
+
1639
+
1640
+
1641
+
1642
+
1643
+
1644
+
1645
+
1646
+
1647
+
1648
+
1649
+
1650
+
1651
+
1652
+
1653
+
1654
+
1655
+
1656
+
1657
+
1658
+
1659
+
1660
+
1661
+
1662
+
1663
+
1664
+
1665
+
1666
+
1667
+
1668
+
1669
+
1670
+
1671
+
1672
+
1673
+
1674
+
1675
+
1676
+
1677
+
1678
+
1679
+
1680
+
1681
+
1682
+
1683
+
1684
+
1685
+
1686
+
1687
+
1688
+
1689
+
1690
+
1691
+
1692
+
1693
+
1694
+
1695
+
1696
+
1697
+
1698
+
1699
+
1700
+
1701
+
1702
+
1703
+
1704
+
1705
+
1706
+
1707
+
1708
+
1709
+
1710
+
1711
+
1712
+
1713
+
1714
+
1715
+
1716
+
1717
+
1718
+
1719
+
1720
+
1721
+
1722
+
1723
+
1724
+
1725
+
1726
+
1727
+
1728
+
1729
+
1730
+
1731
+
1732
+
1733
+
1734
+
1735
+
1736
+
1737
+
1738
+
1739
+
1740
+
1741
+
1742
+
1743
+
1744
+
1745
+
1746
+
1747
+
1748
+
1749
+
1750
+
1751
+
1752
+
1753
+
1754
+
1755
+
1756
+
1757
+
1758
+
1759
+
1760
+
1761
+
1762
+
1763
+
1764
+
1765
+
1766
+
1767
+
1768
+
1769
+
1770
+
1771
+
1772
+
1773
+
1774
+
1775
+
1776
+
1777
+
1778
+
1779
+
1780
+
1781
+
1782
+
1783
+
1784
+
1785
+ Configuration saved in ./checkpoint-500/config.json
1786
+ {'eval_loss': 2.797088623046875, 'eval_wer': 1.0, 'eval_runtime': 2782.5245, 'eval_samples_per_second': 14.641, 'eval_steps_per_second': 1.22, 'epoch': 0.04}
1787
+ Model weights saved in ./checkpoint-500/pytorch_model.bin
1788
+ Feature extractor saved in ./checkpoint-500/preprocessor_config.json
wandb/run-20220523_115145-3dybzmyz/files/requirements.txt ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ aiohttp==3.8.1
2
+ aiosignal==1.2.0
3
+ appdirs==1.4.4
4
+ async-timeout==4.0.2
5
+ attrs==21.4.0
6
+ audioread==2.1.9
7
+ certifi==2021.10.8
8
+ cffi==1.15.0
9
+ charset-normalizer==2.0.12
10
+ click==8.1.2
11
+ datasets==2.1.0
12
+ decorator==5.1.1
13
+ dill==0.3.4
14
+ docker-pycreds==0.4.0
15
+ filelock==3.6.0
16
+ frozenlist==1.3.0
17
+ fsspec==2022.3.0
18
+ gitdb==4.0.9
19
+ gitpython==3.1.27
20
+ huggingface-hub==0.5.1
21
+ hypothesis==6.46.5
22
+ idna==3.3
23
+ jiwer==2.3.0
24
+ joblib==1.1.0
25
+ kenlm==0.0.0
26
+ librosa==0.9.1
27
+ llvmlite==0.38.0
28
+ multidict==6.0.2
29
+ multiprocess==0.70.12.2
30
+ numba==0.55.1
31
+ numpy==1.21.6
32
+ packaging==21.3
33
+ pandas==1.4.2
34
+ pathtools==0.1.2
35
+ pillow==9.1.0
36
+ pip==20.3.4
37
+ pkg-resources==0.0.0
38
+ pooch==1.6.0
39
+ promise==2.3
40
+ protobuf==3.20.1
41
+ psutil==5.9.0
42
+ pyarrow==7.0.0
43
+ pycparser==2.21
44
+ pyctcdecode==0.3.0
45
+ pygtrie==2.4.2
46
+ pyparsing==3.0.8
47
+ python-dateutil==2.8.2
48
+ python-levenshtein==0.12.2
49
+ pytz==2022.1
50
+ pyyaml==6.0
51
+ regex==2022.4.24
52
+ requests==2.27.1
53
+ resampy==0.2.2
54
+ responses==0.18.0
55
+ sacremoses==0.0.49
56
+ scikit-learn==1.0.2
57
+ scipy==1.8.0
58
+ sentry-sdk==1.5.10
59
+ setproctitle==1.2.3
60
+ setuptools==44.1.1
61
+ shortuuid==1.0.8
62
+ six==1.16.0
63
+ smmap==5.0.0
64
+ sortedcontainers==2.4.0
65
+ soundfile==0.10.3.post1
66
+ threadpoolctl==3.1.0
67
+ tokenizers==0.12.1
68
+ torch==1.11.0+cu113
69
+ torchaudio==0.11.0+cu113
70
+ torchvision==0.12.0+cu113
71
+ tqdm==4.64.0
72
+ transformers==4.18.0
73
+ typing-extensions==4.2.0
74
+ urllib3==1.26.9
75
+ wandb==0.12.15
76
+ xxhash==3.0.0
77
+ yarl==1.7.2
wandb/run-20220523_115145-3dybzmyz/files/wandb-metadata.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "os": "Linux-5.13.0-40-generic-x86_64-with-glibc2.34",
3
+ "python": "3.9.7",
4
+ "heartbeatAt": "2022-05-23T09:51:46.677597",
5
+ "startedAt": "2022-05-23T09:51:45.586535",
6
+ "docker": null,
7
+ "cpu_count": 96,
8
+ "cuda": null,
9
+ "args": [
10
+ "--model_name_or_path=facebook/wav2vec2-xls-r-1b",
11
+ "--hub_model_id=NbAiLab/wav2vec2-1b-npsc-nst-bokmaal",
12
+ "--output_dir=./",
13
+ "--overwrite_output_dir",
14
+ "--num_train_epochs=40",
15
+ "--per_device_train_batch_size=12",
16
+ "--per_device_eval_batch_size=12",
17
+ "--gradient_accumulation_steps=2",
18
+ "--learning_rate=2e-5",
19
+ "--warmup_steps=2000",
20
+ "--length_column_name=input_length",
21
+ "--evaluation_strategy=steps",
22
+ "--text_column_name=text",
23
+ "--save_steps=500",
24
+ "--eval_steps=500",
25
+ "--logging_steps=100",
26
+ "--layerdrop=0.041",
27
+ "--attention_dropout=0.094",
28
+ "--activation_dropout=0.055",
29
+ "--hidden_dropout=0.047",
30
+ "--save_total_limit=3",
31
+ "--freeze_feature_encoder",
32
+ "--feat_proj_dropout=0.04",
33
+ "--mask_time_prob=0.082",
34
+ "--mask_time_length=10",
35
+ "--mask_feature_prob=0.25",
36
+ "--mask_feature_length=64",
37
+ "--gradient_checkpointing",
38
+ "--min_duration_in_seconds=0.5",
39
+ "--max_duration_in_seconds=30.0",
40
+ "--use_auth_token",
41
+ "--seed=42",
42
+ "--fp16",
43
+ "--group_by_length",
44
+ "--do_train",
45
+ "--do_eval",
46
+ "--push_to_hub",
47
+ "--preprocessing_num_workers=32",
48
+ "--ctc_zero_infinity"
49
+ ],
50
+ "state": "running",
51
+ "program": "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py",
52
+ "codePath": "run_speech_recognition_ctc.py",
53
+ "git": {
54
+ "remote": "https://huggingface.co/NbAiLab/wav2vec2-1b-npsc-nst-bokmaal",
55
+ "commit": "05ba5dd22c9cec7365fba75fae8630eec1302aff"
56
+ },
57
+ "email": "[email protected]",
58
+ "root": "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal",
59
+ "host": "dante",
60
+ "username": "rolvb",
61
+ "executable": "/mnt/lv_ai_1_dante/ml/rolvb/venv/bin/python"
62
+ }
wandb/run-20220523_115145-3dybzmyz/files/wandb-summary.json ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_115145-3dybzmyz/logs/debug-internal.log ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220523_115145-3dybzmyz/logs/debug.log ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2022-05-23 11:51:45,597 INFO MainThread:1049120 [wandb_setup.py:_flush():75] Loading settings from /home/rolvb/.config/wandb/settings
2
+ 2022-05-23 11:51:45,597 INFO MainThread:1049120 [wandb_setup.py:_flush():75] Loading settings from /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/settings
3
+ 2022-05-23 11:51:45,597 INFO MainThread:1049120 [wandb_setup.py:_flush():75] Loading settings from environment variables: {'project': 'wav2vec2', 'entity': 'NbAiLab'}
4
+ 2022-05-23 11:51:45,597 INFO MainThread:1049120 [wandb_setup.py:_flush():75] Inferring run settings from compute environment: {'program_relpath': 'run_speech_recognition_ctc.py', 'program': '/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/run_speech_recognition_ctc.py'}
5
+ 2022-05-23 11:51:45,597 INFO MainThread:1049120 [wandb_init.py:_log_setup():437] Logging user logs to /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/run-20220523_115145-3dybzmyz/logs/debug.log
6
+ 2022-05-23 11:51:45,597 INFO MainThread:1049120 [wandb_init.py:_log_setup():438] Logging internal logs to /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-npsc-nst-bokmaal/wandb/run-20220523_115145-3dybzmyz/logs/debug-internal.log
7
+ 2022-05-23 11:51:45,598 INFO MainThread:1049120 [wandb_init.py:init():471] calling init triggers
8
+ 2022-05-23 11:51:45,598 INFO MainThread:1049120 [wandb_init.py:init():474] wandb.init called with sweep_config: {}
9
+ config: {}
10
+ 2022-05-23 11:51:45,598 INFO MainThread:1049120 [wandb_init.py:init():524] starting backend
11
+ 2022-05-23 11:51:45,598 INFO MainThread:1049120 [backend.py:_multiprocessing_setup():97] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
12
+ 2022-05-23 11:51:45,689 INFO MainThread:1049120 [backend.py:ensure_launched():217] starting backend process...
13
+ 2022-05-23 11:51:45,775 INFO MainThread:1049120 [backend.py:ensure_launched():222] started backend process with pid: 1049978
14
+ 2022-05-23 11:51:45,776 INFO MainThread:1049120 [wandb_init.py:init():533] backend started and connected
15
+ 2022-05-23 11:51:45,787 INFO MainThread:1049120 [wandb_init.py:init():597] updated telemetry
16
+ 2022-05-23 11:51:45,996 INFO MainThread:1049120 [wandb_init.py:init():628] communicating run to backend with 30 second timeout
17
+ 2022-05-23 11:51:46,532 INFO MainThread:1049120 [wandb_run.py:_on_init():1923] communicating current version
18
+ 2022-05-23 11:51:46,660 INFO MainThread:1049120 [wandb_run.py:_on_init():1927] got version response upgrade_message: "wandb version 0.12.16 is available! To upgrade, please run:\n $ pip install wandb --upgrade"
19
+
20
+ 2022-05-23 11:51:46,660 INFO MainThread:1049120 [wandb_init.py:init():659] starting run threads in backend
21
+ 2022-05-23 11:51:46,714 INFO MainThread:1049120 [wandb_run.py:_console_start():1897] atexit reg
22
+ 2022-05-23 11:51:46,715 INFO MainThread:1049120 [wandb_run.py:_redirect():1770] redirect: SettingsConsole.REDIRECT
23
+ 2022-05-23 11:51:46,716 INFO MainThread:1049120 [wandb_run.py:_redirect():1775] Redirecting console.
24
+ 2022-05-23 11:51:46,718 INFO MainThread:1049120 [wandb_run.py:_redirect():1831] Redirects installed.
25
+ 2022-05-23 11:51:46,719 INFO MainThread:1049120 [wandb_init.py:init():684] run started, returning control to user process
26
+ 2022-05-23 11:51:46,741 INFO MainThread:1049120 [wandb_run.py:_config_callback():1131] config_cb None None {'return_dict': True, 'output_hidden_states': False, 'output_attentions': False, 'torchscript': False, 'torch_dtype': 'float32', 'use_bfloat16': False, 'pruned_heads': {}, 'tie_word_embeddings': True, 'is_encoder_decoder': False, 'is_decoder': False, 'cross_attention_hidden_size': None, 'add_cross_attention': False, 'tie_encoder_decoder': False, 'max_length': 20, 'min_length': 0, 'do_sample': False, 'early_stopping': False, 'num_beams': 1, 'num_beam_groups': 1, 'diversity_penalty': 0.0, 'temperature': 1.0, 'top_k': 50, 'top_p': 1.0, 'typical_p': 1.0, 'repetition_penalty': 1.0, 'length_penalty': 1.0, 'no_repeat_ngram_size': 0, 'encoder_no_repeat_ngram_size': 0, 'bad_words_ids': None, 'num_return_sequences': 1, 'chunk_size_feed_forward': 0, 'output_scores': False, 'return_dict_in_generate': False, 'forced_bos_token_id': None, 'forced_eos_token_id': None, 'remove_invalid_values': False, 'exponential_decay_length_penalty': None, 'architectures': ['Wav2Vec2ForPreTraining'], 'finetuning_task': None, 'id2label': {0: 'LABEL_0', 1: 'LABEL_1'}, 'label2id': {'LABEL_0': 0, 'LABEL_1': 1}, 'tokenizer_class': None, 'prefix': None, 'bos_token_id': 1, 'pad_token_id': 31, 'eos_token_id': 2, 'sep_token_id': None, 'decoder_start_token_id': None, 'task_specific_params': None, 'problem_type': None, '_name_or_path': 'facebook/wav2vec2-xls-r-1b', 'transformers_version': '4.18.0', 'feat_extract_dropout': 0.0, 'model_type': 'wav2vec2', 'num_feat_extract_layers': 7, 'hidden_size': 1280, 'feat_extract_norm': 'layer', 'feat_extract_activation': 'gelu', 'conv_dim': [512, 512, 512, 512, 512, 512, 512], 'conv_stride': [5, 2, 2, 2, 2, 2, 2], 'conv_kernel': [10, 3, 3, 3, 3, 2, 2], 'conv_bias': True, 'num_conv_pos_embeddings': 128, 'num_conv_pos_embedding_groups': 16, 'num_hidden_layers': 48, 'intermediate_size': 5120, 'hidden_act': 'gelu', 'num_attention_heads': 16, 'hidden_dropout': 0.047, 'attention_dropout': 0.094, 'activation_dropout': 0.055, 'feat_proj_dropout': 0.04, 'final_dropout': 0.0, 'layerdrop': 0.041, 'layer_norm_eps': 1e-05, 'initializer_range': 0.02, 'vocab_size': 34, 'do_stable_layer_norm': True, 'use_weighted_layer_sum': False, 'apply_spec_augment': True, 'mask_time_prob': 0.082, 'mask_time_length': 10, 'mask_time_min_masks': 2, 'mask_feature_prob': 0.25, 'mask_feature_length': 64, 'mask_feature_min_masks': 0, 'num_codevectors_per_group': 320, 'num_codevector_groups': 2, 'contrastive_logits_temperature': 0.1, 'feat_quantizer_dropout': 0.0, 'num_negatives': 100, 'codevector_dim': 1024, 'proj_codevector_dim': 1024, 'diversity_loss_weight': 0.1, 'ctc_loss_reduction': 'mean', 'ctc_zero_infinity': True, 'add_adapter': False, 'adapter_kernel_size': 3, 'adapter_stride': 2, 'num_adapter_layers': 3, 'output_hidden_size': 1280, 'classifier_proj_size': 256, 'tdnn_dim': [512, 512, 512, 512, 1500], 'tdnn_kernel': [5, 3, 3, 1, 1], 'tdnn_dilation': [1, 2, 3, 1, 1], 'xvector_output_dim': 512, 'output_dir': './', 'overwrite_output_dir': True, 'do_train': True, 'do_eval': True, 'do_predict': False, 'evaluation_strategy': 'steps', 'prediction_loss_only': False, 'per_device_train_batch_size': 12, 'per_device_eval_batch_size': 12, 'per_gpu_train_batch_size': 'None', 'per_gpu_eval_batch_size': 'None', 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': 'None', 'eval_delay': 0, 'learning_rate': 2e-05, 'weight_decay': 0.0, 'adam_beta1': 0.9, 'adam_beta2': 0.999, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 40.0, 'max_steps': -1, 'lr_scheduler_type': 'linear', 'warmup_ratio': 0.0, 'warmup_steps': 2000, 'log_level': -1, 'log_level_replica': -1, 'log_on_each_node': True, 'logging_dir': './runs/May23_11-33-45_dante', 'logging_strategy': 'steps', 'logging_first_step': False, 'logging_steps': 100, 'logging_nan_inf_filter': True, 'save_strategy': 'steps', 'save_steps': 500, 'save_total_limit': 3, 'save_on_each_node': False, 'no_cuda': False, 'seed': 42, 'data_seed': 'None', 'bf16': False, 'fp16': True, 'fp16_opt_level': 'O1', 'half_precision_backend': 'amp', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': 'None', 'local_rank': -1, 'xpu_backend': 'None', 'tpu_num_cores': 'None', 'tpu_metrics_debug': False, 'debug': '[]', 'dataloader_drop_last': False, 'eval_steps': 500, 'dataloader_num_workers': 0, 'past_index': -1, 'run_name': './', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': 'None', 'load_best_model_at_end': False, 'metric_for_best_model': 'None', 'greater_is_better': 'None', 'ignore_data_skip': False, 'sharded_ddp': '[]', 'deepspeed': 'None', 'label_smoothing_factor': 0.0, 'optim': 'adamw_hf', 'adafactor': False, 'group_by_length': True, 'length_column_name': 'input_length', 'report_to': "['wandb']", 'ddp_find_unused_parameters': 'None', 'ddp_bucket_cap_mb': 'None', 'dataloader_pin_memory': True, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': True, 'resume_from_checkpoint': 'None', 'hub_model_id': 'NbAiLab/wav2vec2-1b-npsc-nst-bokmaal', 'hub_strategy': 'every_save', 'hub_token': '<HUB_TOKEN>', 'gradient_checkpointing': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': 'None', 'push_to_hub_organization': 'None', 'push_to_hub_token': '<PUSH_TO_HUB_TOKEN>', '_n_gpu': 1, 'mp_parameters': '', 'train_batch_size': 12, 'eval_batch_size': 12}
27
+ 2022-05-23 11:51:46,744 INFO MainThread:1049120 [wandb_watch.py:watch():47] Watching
wandb/run-20220523_115145-3dybzmyz/run-3dybzmyz.wandb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c91f77b21fe4c893f5ed1d5a60bb91ac15b749c4c7e6440f692d28c0673f002
3
+ size 7716664