| model: iic/SenseVoiceSmall | |
| trust_remote_code: true | |
| train_data_set_list: /root/autodl-tmp/train/train.jsonl | |
| valid_data_set_list: /root/autodl-tmp/val/val.jsonl | |
| dataset_conf: | |
| data_split_num: 1 | |
| batch_sampler: BatchSampler | |
| batch_size: 25000 | |
| sort_size: 1024 | |
| batch_type: token | |
| num_workers: 4 | |
| train_conf: | |
| max_epoch: 100 | |
| log_interval: 1 | |
| resume: true | |
| validate_interval: 2000 | |
| save_checkpoint_interval: 2000 | |
| keep_nbest_models: 10 | |
| avg_nbest_model: 10 | |
| use_deepspeed: false | |
| deepspeed_config: /root/autodl-tmp/SenseVoice/deepspeed_conf/ds_stage1.json | |
| optim_conf: | |
| lr: 0.0002 | |
| output_dir: ./outputs | |