Kevin Fink
commited on
Commit
·
1e30251
1
Parent(s):
ee9996e
dev
Browse files
app.py
CHANGED
@@ -151,7 +151,7 @@ def fine_tune_model(model, dataset_name, hub_id, api_key, num_epochs, batch_size
|
|
151 |
print(dataset)
|
152 |
print(dataset.keys())
|
153 |
#tokenized_second_half = tokenize_function(third_third)
|
154 |
-
tokenized_second_half = dataset.map(tokenize_function)
|
155 |
dataset['train'] = concatenate_datasets([saved_dataset['train'], tokenized_second_half['train']])
|
156 |
dataset['train'].save_to_disk(f'/data/{hub_id.strip()}_train_dataset3')
|
157 |
return 'THIRD THIRD LOADED'
|
|
|
151 |
print(dataset)
|
152 |
print(dataset.keys())
|
153 |
#tokenized_second_half = tokenize_function(third_third)
|
154 |
+
tokenized_second_half = dataset.map(tokenize_function, batched=True)
|
155 |
dataset['train'] = concatenate_datasets([saved_dataset['train'], tokenized_second_half['train']])
|
156 |
dataset['train'].save_to_disk(f'/data/{hub_id.strip()}_train_dataset3')
|
157 |
return 'THIRD THIRD LOADED'
|