init

training_scripts/finetune_t5.py

@@ -85,7 +85,7 @@ def train(
         skip_train: bool = False,
         skip_test: bool = False,
         skip_upload: bool = False,
-        eval_batch_size: int = None):
+        batch_eval: int = None):
     """Fine-tune seq2seq model."""
     logging.info(f'[CONFIG]\n\t *LM: {model_name}, \n\t *Data: {dataset} ({dataset_name})')
     if not output_dir:
@@ -139,7 +139,7 @@ def train(
         lr = [1e-6, 1e-5, 1e-4] if lr is None else lr
         batch = [32] if not batch else batch
         epoch = [3, 5] if not epoch else epoch
-        eval_batch_size = min(batch) if not eval_batch_size else eval_batch_size
+        batch_eval = min(batch) if not batch_eval else batch_eval
         for n, (lr_tmp, batch_tmp, epoch_tmp) in enumerate(product(lr, batch, epoch)):
             logging.info(f"[TRAIN {n}/{len(lr) * len(batch) * len(epoch)}] lr: {lr_tmp}, batch: {batch_tmp}")
             output_dir_tmp = f"{output_dir}/model_lr_{lr_tmp}_batch_{batch_tmp}_epoch_{epoch_tmp}"
@@ -155,7 +155,7 @@ def train(
                     learning_rate=lr_tmp,
                     output_dir=output_dir_tmp,
                     evaluation_strategy="no",
-                    per_device_eval_batch_size=eval_batch_size,
+                    per_device_eval_batch_size=batch_eval,
                     seed=random_seed,
                     per_device_train_batch_size=batch_tmp,
                 ),
@@ -209,7 +209,7 @@ def train(
                 device='cuda:0' if torch.cuda.is_available() else 'cpu',
             )
             input_data = [i[dataset_column_text] for i in dataset_instance[dataset_split_test]]
-            output = pipe(input_data, batch_size=eval_batch_size)
+            output = pipe(input_data, batch_size=batch_eval)
             output = [i['generated_text'] for i in output]
             with open(f'{output_dir}/best_model/prediction_test.txt', 'w') as f:
                 f.write('\n'.join(output))
@@ -281,6 +281,7 @@ if __name__ == '__main__':
     parser.add_argument('--lr', nargs='+', default=None, type=float)
     parser.add_argument('--epoch', nargs='+', default=None, type=int)
     parser.add_argument('--batch', nargs='+', default=None, type=int)
+    parser.add_argument('--batch-eval', type=int, default=None)
     parser.add_argument('--down-sample-train', default=None, type=int)
     parser.add_argument('--down-sample-validation', default=2000, type=int)
     parser.add_argument('--random-seed', default=42, type=int)
@@ -307,6 +308,7 @@ if __name__ == '__main__':
         lr=opt.lr,
         epoch=opt.epoch,
         batch=opt.batch,
+        batch_eval=opt.batch_eval,
         down_sample_train=opt.down_sample_train,
         down_sample_validation=opt.down_sample_validation,
         random_seed=opt.random_seed,