File size: 1,354 Bytes
7b6e528
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
python run_speech_recognition_seq2seq_streaming.py \
--model_name_or_path="openai/whisper-medium" \
--dataset_name="language-and-voice-lab/samromur_asr+language-and-voice-lab/althingi_asr+language-and-voice-lab/malromur_asr+language-and-voice-lab/samromur_children+language-and-voice-lab/raddromur_asr" \
--language="icelandic" \
--train_split_name="train" \
--eval_dataset_name="language-and-voice-lab/samromur_asr" \
--eval_split_name="validation" \
--eval_steps="500" \
--max_eval_samples="1000" \
--model_index_name="Whisper medium Icelandic" \
--max_steps="10000" \
--output_dir="./" \
--per_device_train_batch_size="32" \
--gradient_accumulation_steps="2" \
--per_device_eval_batch_size="32" \
--logging_steps="25" \
--learning_rate="1e-5" \
--warmup_steps="500" \
--evaluation_strategy="steps" \
--save_strategy="steps" \
--save_steps="1000" \
--generation_max_length="225" \
--length_column_name="input_length" \
--max_duration_in_seconds="30" \
--text_column_name="normalized_text" \
--freeze_feature_encoder="False" \
--report_to="tensorboard" \
--metric_for_best_model="wer" \
--greater_is_better="False" \
--load_best_model_at_end \
--gradient_checkpointing \
--fp16 \
--overwrite_output_dir \
--do_train \
--do_eval \
--predict_with_generate \
--do_normalize_eval \
--streaming \
--push_to_hub \
--mapper="/home/ubuntu/whisper/text_mapping"