AndrewMcDowell
commited on
Commit
β’
6a0b76e
1
Parent(s):
17227ca
Training in progress, step 500
Browse files- .ipynb_checkpoints/run-checkpoint.sh +6 -4
- previous_runs/.ipynb_checkpoints/run-checkpoint.sh +35 -0
- previous_runs/run.sh +35 -0
- pytorch_model.bin +1 -1
- run.sh +6 -4
- special_tokens_map.json +1 -1
- training_args.bin +1 -1
.ipynb_checkpoints/run-checkpoint.sh
CHANGED
@@ -3,11 +3,12 @@ python run_speech_recognition_ctc_bnb.py \
|
|
3 |
--model_name_or_path="facebook/wav2vec2-xls-r-1b" \
|
4 |
--dataset_config_name="ar" \
|
5 |
--output_dir="./" \
|
6 |
-
--
|
7 |
-
--
|
|
|
8 |
--per_device_eval_batch_size="8" \
|
9 |
--gradient_accumulation_steps="4" \
|
10 |
-
--learning_rate="
|
11 |
--warmup_steps="2000" \
|
12 |
--length_column_name="input_length" \
|
13 |
--evaluation_strategy="steps" \
|
@@ -30,4 +31,5 @@ python run_speech_recognition_ctc_bnb.py \
|
|
30 |
--fp16 \
|
31 |
--group_by_length \
|
32 |
--do_train --do_eval \
|
33 |
-
--push_to_hub
|
|
|
|
3 |
--model_name_or_path="facebook/wav2vec2-xls-r-1b" \
|
4 |
--dataset_config_name="ar" \
|
5 |
--output_dir="./" \
|
6 |
+
--overwrite_output_dir \
|
7 |
+
--num_train_epochs="10" \
|
8 |
+
--per_device_train_batch_size="16" \
|
9 |
--per_device_eval_batch_size="8" \
|
10 |
--gradient_accumulation_steps="4" \
|
11 |
+
--learning_rate="6.5e-5" \
|
12 |
--warmup_steps="2000" \
|
13 |
--length_column_name="input_length" \
|
14 |
--evaluation_strategy="steps" \
|
|
|
31 |
--fp16 \
|
32 |
--group_by_length \
|
33 |
--do_train --do_eval \
|
34 |
+
--push_to_hub
|
35 |
+
|
previous_runs/.ipynb_checkpoints/run-checkpoint.sh
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
python run_speech_recognition_ctc_bnb.py \
|
2 |
+
--dataset_name="mozilla-foundation/common_voice_8_0" \
|
3 |
+
--model_name_or_path="facebook/wav2vec2-xls-r-1b" \
|
4 |
+
--dataset_config_name="ar" \
|
5 |
+
--output_dir="./" \
|
6 |
+
--num_train_epochs="20" \
|
7 |
+
--per_device_train_batch_size="32" \
|
8 |
+
--per_device_eval_batch_size="8" \
|
9 |
+
--gradient_accumulation_steps="4" \
|
10 |
+
--learning_rate="1.5e-4" \
|
11 |
+
--warmup_steps="2000" \
|
12 |
+
--length_column_name="input_length" \
|
13 |
+
--evaluation_strategy="steps" \
|
14 |
+
--text_column_name="sentence" \
|
15 |
+
--chars_to_ignore , ? . ! \- \; \: \" β % β β οΏ½ β β β¦ β \
|
16 |
+
--save_steps="500" \
|
17 |
+
--eval_steps="500" \
|
18 |
+
--logging_steps="100" \
|
19 |
+
--layerdrop="0.0" \
|
20 |
+
--activation_dropout="0.1" \
|
21 |
+
--save_total_limit="4" \
|
22 |
+
--freeze_feature_encoder \
|
23 |
+
--feat_proj_dropout="0.0" \
|
24 |
+
--mask_time_prob="0.75" \
|
25 |
+
--mask_time_length="10" \
|
26 |
+
--mask_feature_prob="0.25" \
|
27 |
+
--mask_feature_length="64" \
|
28 |
+
--gradient_checkpointing \
|
29 |
+
--use_auth_token \
|
30 |
+
--fp16 \
|
31 |
+
--group_by_length \
|
32 |
+
--do_train --do_eval \
|
33 |
+
--push_to_hub
|
34 |
+
|
35 |
+
# Poor performance. Stopped at {'eval_loss': 1.338714361190796, 'eval_wer': 0.897404591449766, 'eval_runtime': 422.6845, 'eval_samples_per_second': 24.576, 'eval_steps_per_second': 3.073, 'epoch': 10.07}
|
previous_runs/run.sh
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
python run_speech_recognition_ctc_bnb.py \
|
2 |
+
--dataset_name="mozilla-foundation/common_voice_8_0" \
|
3 |
+
--model_name_or_path="facebook/wav2vec2-xls-r-1b" \
|
4 |
+
--dataset_config_name="ar" \
|
5 |
+
--output_dir="./" \
|
6 |
+
--num_train_epochs="20" \
|
7 |
+
--per_device_train_batch_size="32" \
|
8 |
+
--per_device_eval_batch_size="8" \
|
9 |
+
--gradient_accumulation_steps="4" \
|
10 |
+
--learning_rate="1.5e-4" \
|
11 |
+
--warmup_steps="2000" \
|
12 |
+
--length_column_name="input_length" \
|
13 |
+
--evaluation_strategy="steps" \
|
14 |
+
--text_column_name="sentence" \
|
15 |
+
--chars_to_ignore , ? . ! \- \; \: \" β % β β οΏ½ β β β¦ β \
|
16 |
+
--save_steps="500" \
|
17 |
+
--eval_steps="500" \
|
18 |
+
--logging_steps="100" \
|
19 |
+
--layerdrop="0.0" \
|
20 |
+
--activation_dropout="0.1" \
|
21 |
+
--save_total_limit="4" \
|
22 |
+
--freeze_feature_encoder \
|
23 |
+
--feat_proj_dropout="0.0" \
|
24 |
+
--mask_time_prob="0.75" \
|
25 |
+
--mask_time_length="10" \
|
26 |
+
--mask_feature_prob="0.25" \
|
27 |
+
--mask_feature_length="64" \
|
28 |
+
--gradient_checkpointing \
|
29 |
+
--use_auth_token \
|
30 |
+
--fp16 \
|
31 |
+
--group_by_length \
|
32 |
+
--do_train --do_eval \
|
33 |
+
--push_to_hub
|
34 |
+
|
35 |
+
# Poor performance. Stopped at {'eval_loss': 1.338714361190796, 'eval_wer': 0.897404591449766, 'eval_runtime': 422.6845, 'eval_samples_per_second': 24.576, 'eval_steps_per_second': 3.073, 'epoch': 10.07}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3850799473
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:72110b18b9f181a1796af0610a022c116fbc3b53316387df30a797bd09e691ad
|
3 |
size 3850799473
|
run.sh
CHANGED
@@ -3,11 +3,12 @@ python run_speech_recognition_ctc_bnb.py \
|
|
3 |
--model_name_or_path="facebook/wav2vec2-xls-r-1b" \
|
4 |
--dataset_config_name="ar" \
|
5 |
--output_dir="./" \
|
6 |
-
--
|
7 |
-
--
|
|
|
8 |
--per_device_eval_batch_size="8" \
|
9 |
--gradient_accumulation_steps="4" \
|
10 |
-
--learning_rate="
|
11 |
--warmup_steps="2000" \
|
12 |
--length_column_name="input_length" \
|
13 |
--evaluation_strategy="steps" \
|
@@ -30,4 +31,5 @@ python run_speech_recognition_ctc_bnb.py \
|
|
30 |
--fp16 \
|
31 |
--group_by_length \
|
32 |
--do_train --do_eval \
|
33 |
-
--push_to_hub
|
|
|
|
3 |
--model_name_or_path="facebook/wav2vec2-xls-r-1b" \
|
4 |
--dataset_config_name="ar" \
|
5 |
--output_dir="./" \
|
6 |
+
--overwrite_output_dir \
|
7 |
+
--num_train_epochs="10" \
|
8 |
+
--per_device_train_batch_size="16" \
|
9 |
--per_device_eval_batch_size="8" \
|
10 |
--gradient_accumulation_steps="4" \
|
11 |
+
--learning_rate="6.5e-5" \
|
12 |
--warmup_steps="2000" \
|
13 |
--length_column_name="input_length" \
|
14 |
--evaluation_strategy="steps" \
|
|
|
31 |
--fp16 \
|
32 |
--group_by_length \
|
33 |
--do_train --do_eval \
|
34 |
+
--push_to_hub
|
35 |
+
|
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}, {"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}]}
|
|
|
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}, {"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}]}
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2991
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dd4fb4cf661cbec66ff39edfd085a48937505029ebdfd04e974b45d65bbc7c64
|
3 |
size 2991
|