Lint tokenization
Browse files
src/axolotl/utils/tokenization.py
CHANGED
|
@@ -1,5 +1,8 @@
|
|
| 1 |
-
|
|
|
|
|
|
|
| 2 |
import logging
|
|
|
|
| 3 |
|
| 4 |
|
| 5 |
def check_dataset_labels(dataset, tokenizer):
|
|
@@ -17,7 +20,7 @@ def check_example_labels(example, tokenizer):
|
|
| 17 |
# You can compare the input_ids and labels element-wise
|
| 18 |
# Remember to ignore positions with IGNORE_TOKEN_ID (if you use it) or attention_mask equal to 0
|
| 19 |
colored_tokens = []
|
| 20 |
-
for
|
| 21 |
zip(input_ids, labels, attention_mask)
|
| 22 |
):
|
| 23 |
decoded_input_token = tokenizer.decode(input_id)
|
|
|
|
| 1 |
+
"""Module for tokenization utilities"""
|
| 2 |
+
|
| 3 |
+
|
| 4 |
import logging
|
| 5 |
+
from termcolor import colored
|
| 6 |
|
| 7 |
|
| 8 |
def check_dataset_labels(dataset, tokenizer):
|
|
|
|
| 20 |
# You can compare the input_ids and labels element-wise
|
| 21 |
# Remember to ignore positions with IGNORE_TOKEN_ID (if you use it) or attention_mask equal to 0
|
| 22 |
colored_tokens = []
|
| 23 |
+
for _, (input_id, label_id, mask) in enumerate(
|
| 24 |
zip(input_ids, labels, attention_mask)
|
| 25 |
):
|
| 26 |
decoded_input_token = tokenizer.decode(input_id)
|