Training in progress, step 500
Browse files- config.json +2 -2
- pytorch_model.bin +1 -1
- run.sh +2 -2
- run_speech_recognition_ctc.py +5 -2
- runs/Feb01_12-24-21_job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba/1643718355.8126762/events.out.tfevents.1643718355.job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba.2341977.1 +3 -0
- runs/Feb01_12-24-21_job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba/events.out.tfevents.1643718355.job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba.2341977.0 +3 -0
- special_tokens_map.json +1 -1
- training_args.bin +1 -1
config.json
CHANGED
@@ -42,7 +42,7 @@
|
|
42 |
2
|
43 |
],
|
44 |
"ctc_loss_reduction": "mean",
|
45 |
-
"ctc_zero_infinity":
|
46 |
"diversity_loss_weight": 0.1,
|
47 |
"do_stable_layer_norm": true,
|
48 |
"eos_token_id": 2,
|
@@ -58,7 +58,7 @@
|
|
58 |
"initializer_range": 0.02,
|
59 |
"intermediate_size": 4096,
|
60 |
"layer_norm_eps": 1e-05,
|
61 |
-
"layerdrop": 0.
|
62 |
"mask_feature_length": 64,
|
63 |
"mask_feature_min_masks": 0,
|
64 |
"mask_feature_prob": 0.25,
|
|
|
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,
|
|
|
58 |
"initializer_range": 0.02,
|
59 |
"intermediate_size": 4096,
|
60 |
"layer_norm_eps": 1e-05,
|
61 |
+
"layerdrop": 0.1,
|
62 |
"mask_feature_length": 64,
|
63 |
"mask_feature_min_masks": 0,
|
64 |
"mask_feature_prob": 0.25,
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1262067185
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:801ed6a8fee74d72d6fc0685e04cd256f1e03e216b971c82295eba8597bb5a44
|
3 |
size 1262067185
|
run.sh
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
python run_speech_recognition_ctc.py \
|
2 |
--dataset_name="NbAiLab/NPSC" \
|
3 |
--model_name_or_path="facebook/wav2vec2-xls-r-300m" \
|
4 |
-
|
5 |
--dataset_config_name="16K_mp3" \
|
6 |
--output_dir="./" \
|
7 |
--overwrite_output_dir \
|
@@ -17,7 +17,7 @@ python run_speech_recognition_ctc.py \
|
|
17 |
--save_steps="500" \
|
18 |
--eval_steps="500" \
|
19 |
--logging_steps="100" \
|
20 |
-
--layerdrop="0.
|
21 |
--activation_dropout="0.1" \
|
22 |
--save_total_limit="3" \
|
23 |
--freeze_feature_encoder \
|
|
|
1 |
python run_speech_recognition_ctc.py \
|
2 |
--dataset_name="NbAiLab/NPSC" \
|
3 |
--model_name_or_path="facebook/wav2vec2-xls-r-300m" \
|
4 |
+
--hub_model_id="NbAiLab/wav2vec2-xlsr-300M-NPSC-OH" \
|
5 |
--dataset_config_name="16K_mp3" \
|
6 |
--output_dir="./" \
|
7 |
--overwrite_output_dir \
|
|
|
17 |
--save_steps="500" \
|
18 |
--eval_steps="500" \
|
19 |
--logging_steps="100" \
|
20 |
+
--layerdrop="0.1" \
|
21 |
--activation_dropout="0.1" \
|
22 |
--save_total_limit="3" \
|
23 |
--freeze_feature_encoder \
|
run_speech_recognition_ctc.py
CHANGED
@@ -400,8 +400,10 @@ def main():
|
|
400 |
return re.search("nb-no", entry["sentence_language_code"], flags=re.IGNORECASE)
|
401 |
|
402 |
def filter_tooshort(entry):
|
403 |
-
|
404 |
-
|
|
|
|
|
405 |
|
406 |
def map_dataset(entry):
|
407 |
batch = {"text": entry["text"].lower()}
|
@@ -570,6 +572,7 @@ def main():
|
|
570 |
"gradient_checkpointing": training_args.gradient_checkpointing,
|
571 |
"layerdrop": model_args.layerdrop,
|
572 |
"ctc_loss_reduction": model_args.ctc_loss_reduction,
|
|
|
573 |
"pad_token_id": tokenizer.pad_token_id,
|
574 |
"vocab_size": len(tokenizer),
|
575 |
"activation_dropout": model_args.activation_dropout,
|
|
|
400 |
return re.search("nb-no", entry["sentence_language_code"], flags=re.IGNORECASE)
|
401 |
|
402 |
def filter_tooshort(entry):
|
403 |
+
if len(entry["audio"]["array"]) // 320 < len(entry["text"]):
|
404 |
+
print(f"The audio sample ({entry['audio']['path']}) is too small, and has been omitted. ")
|
405 |
+
return False
|
406 |
+
return True
|
407 |
|
408 |
def map_dataset(entry):
|
409 |
batch = {"text": entry["text"].lower()}
|
|
|
572 |
"gradient_checkpointing": training_args.gradient_checkpointing,
|
573 |
"layerdrop": model_args.layerdrop,
|
574 |
"ctc_loss_reduction": model_args.ctc_loss_reduction,
|
575 |
+
"ctc_zero_infinity": True,
|
576 |
"pad_token_id": tokenizer.pad_token_id,
|
577 |
"vocab_size": len(tokenizer),
|
578 |
"activation_dropout": model_args.activation_dropout,
|
runs/Feb01_12-24-21_job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba/1643718355.8126762/events.out.tfevents.1643718355.job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba.2341977.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:88d8214f159fbe7516815824430a8b5063ea85a2c81ec9b3aa643edb48f424cf
|
3 |
+
size 4783
|
runs/Feb01_12-24-21_job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba/events.out.tfevents.1643718355.job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba.2341977.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c958c822ada84b1247c0379441b9259329a8e242947ae3b15492c8bdbf124b2e
|
3 |
+
size 5815
|
special_tokens_map.json
CHANGED
@@ -1 +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}
|
|
|
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}]}
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3055
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:642fc6041127b9c77f68ff427a0324d8a66f227388841dc1c5a1070b6b9fabb9
|
3 |
size 3055
|