| command: | |
| - python3 | |
| - ${program} | |
| - --overwrite_output_dir | |
| - --freeze_feature_encoder | |
| - --predict_with_generate | |
| - --do_lower_case | |
| - --do_train | |
| - --do_eval | |
| - ${args} | |
| method: random | |
| metric: | |
| goal: minimize | |
| name: eval/wer | |
| parameters: | |
| dataset_cache_dir: | |
| value: /home/sanchitgandhi/cache/huggingface/datasets | |
| dataset_config_name: | |
| value: clean | |
| dataset_name: | |
| value: librispeech_asr | |
| eval_split_name: | |
| value: validation | |
| eval_steps: | |
| value: 500 | |
| generation_max_length: | |
| value: 40 | |
| generation_num_beams: | |
| value: 1 | |
| gradient_accumulation_steps: | |
| value: 1 | |
| gradient_checkpointing: | |
| value: True | |
| learning_rate: | |
| distribution: log_uniform | |
| max: -6.9 | |
| min: -9.2 | |
| length_column_name: | |
| value: input_length | |
| logging_steps: | |
| value: 25 | |
| max_duration_in_seconds: | |
| value: 20 | |
| max_target_length: | |
| value: 128 | |
| model_name_or_path: | |
| value: ./ | |
| num_train_epochs: | |
| value: 3 | |
| output_dir: | |
| value: ./output_dir | |
| per_device_eval_batch_size: | |
| value: 2 | |
| per_device_train_batch_size: | |
| value: 1 | |
| preprocessing_num_workers: | |
| value: 16 | |
| text_column_name: | |
| value: text | |
| train_split_name: | |
| value: train.100 | |
| warmup_steps: | |
| value: 500 | |
| program: run_flax_speech_recognition_seq2seq.py | |
| project: flax-wav2vec2-2-bart-large-checkpointing-scan | |