| | |
|---|
| | | train_conf: |
|---|
| | | accum_grad: 1 |
|---|
| | | grad_clip: 5 |
|---|
| | | max_epoch: 150 |
|---|
| | | max_epoch: 15 |
|---|
| | | keep_nbest_models: 10 |
|---|
| | | log_interval: 10 |
|---|
| | | |
|---|
| | | optim: adamw |
|---|
| | | optim_conf: |
|---|
| | | lr: 0.0001 |
|---|
| | | weight_decay: 0.000001 |
|---|
| | | weight_decay: 0 |
|---|
| | | |
|---|
| | | scheduler: warmuplr |
|---|
| | | scheduler_conf: |
|---|
| | | warmup_steps: 1500 |
|---|
| | |
|---|
| | | dataset: AudioLLMVicunaDataset |
|---|
| | | dataset_conf: |
|---|
| | | index_ds: IndexDSJsonl |
|---|
| | | batch_sampler: RankFullLocalShuffleBatchSampler |
|---|
| | | batch_sampler: CustomDistributedBatchSampler |
|---|
| | | batch_type: example # example or length |
|---|
| | | batch_size: 8 # if batch_type is example, batch_size is the numbers of samples; if length, batch_size is source_token_len+target_token_len; |
|---|
| | | max_token_length: 2048 # filter samples if source_token_len+target_token_len > max_token_length, |
|---|
| | | buffer_size: 500 |
|---|
| | | batch_size: 4 # if batch_type is example, batch_size is the numbers of samples; if length, batch_size is source_token_len+target_token_len; |
|---|
| | | max_token_length: 3000 # filter samples if source_token_len+target_token_len > max_token_length, |
|---|
| | | shuffle: True |
|---|
| | | num_workers: 4 |
|---|
| | | # preprocessor_text: TextPreprocessRemovePunctuation |
|---|
