flpelerin
commited on
Commit
·
87a4d20
1
Parent(s):
29a6938
fix
Browse files
train.py
CHANGED
|
@@ -44,7 +44,7 @@ def process_function(examples):
|
|
| 44 |
examples['text'],
|
| 45 |
padding='max_length', # Fixed padding
|
| 46 |
truncation=True,
|
| 47 |
-
max_length=seq_length # Fixed max length
|
| 48 |
)
|
| 49 |
|
| 50 |
tokenized_datasets = dataset.map(process_function, batched=True)
|
|
|
|
| 44 |
examples['text'],
|
| 45 |
padding='max_length', # Fixed padding
|
| 46 |
truncation=True,
|
| 47 |
+
max_length=(seq_length * batch_size) # Fixed max length
|
| 48 |
)
|
| 49 |
|
| 50 |
tokenized_datasets = dataset.map(process_function, batched=True)
|