Kevin Fink commited on
Commit
ae2e833
·
1 Parent(s): baa8d35
Files changed (1) hide show
  1. app.py +1 -0
app.py CHANGED
@@ -148,6 +148,7 @@ def fine_tune_model(model, dataset_name, hub_id, api_key, num_epochs, batch_size
148
  third_third = dataset['train'].select(range(third_size*2, train_size))
149
  dataset['train'] = third_third
150
  del dataset['test']
 
151
  tokenized_second_half = dataset.map(tokenize_function, batched=True)
152
  dataset['train'] = concatenate_datasets([saved_dataset['train'], tokenized_second_half['train']])
153
  dataset['train'].save_to_disk(f'/data/{hub_id.strip()}_train_dataset3')
 
148
  third_third = dataset['train'].select(range(third_size*2, train_size))
149
  dataset['train'] = third_third
150
  del dataset['test']
151
+ del dataset['validation']
152
  tokenized_second_half = dataset.map(tokenize_function, batched=True)
153
  dataset['train'] = concatenate_datasets([saved_dataset['train'], tokenized_second_half['train']])
154
  dataset['train'].save_to_disk(f'/data/{hub_id.strip()}_train_dataset3')