| 108 | return eval_results |
| 109 | |
| 110 | def parse_args(): |
| 111 | parser = argparse.ArgumentParser(description='Finetune T5 model') |
| 112 | |
| 113 | parser.add_argument('--training-file', dest='training_file', required=False, help='Path to training file', |
| 114 | default=None) |
| 115 | parser.add_argument('--noisy-file', dest='noisy_file', required=False, help='Path to noisy file', |
| 116 | default=None) |
| 117 | parser.add_argument('--validation-file', dest='validation_file', required=False, help='Path to validation file') |
| 118 | parser.add_argument('--language-model', default='t5-base', help='Can be either some huggingface model or a ' |
| 119 | 'path to a model. If the path is in GCS we ' |
| 120 | 'download it first.') |
| 121 | parser.add_argument('--model-dir', dest='model_dir', required=True, |
| 122 | help='the folder/google bucket in which the model will be stored or loaded from.') |
| 123 | parser.add_argument('--epochs', default=20, |
| 124 | help='number of epochs to train') |
| 125 | parser.add_argument('--batch-size', default=4, |
| 126 | help='batch size') |
| 127 | parser.add_argument('--val-batch-size', default=4, |
| 128 | help='validation batch size') |
| 129 | parser.add_argument('--lr', default=0.0001, |
| 130 | help='learning rate') |
| 131 | parser.add_argument('--seq_len', default=256, |
| 132 | help='sequence length') |
| 133 | parser.add_argument('--gradient-accumulation', default=4) |
| 134 | parser.add_argument('--local_rank', default=-1) |
| 135 | args = parser.parse_args() |
| 136 | |
| 137 | return args |
| 138 | |
| 139 | |
| 140 | def main(): |