Hezar: Upload tokenizer_config.yaml
Browse files
preprocessor/tokenizer_config.yaml
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: tokenizer
|
| 2 |
+
config_type: preprocessor
|
| 3 |
+
model: wordpiece
|
| 4 |
+
max_length: 512
|
| 5 |
+
truncation_strategy: longest_first
|
| 6 |
+
truncation_direction: right
|
| 7 |
+
stride: 0
|
| 8 |
+
padding_strategy: longest
|
| 9 |
+
padding_direction: right
|
| 10 |
+
pad_token_id: 0
|
| 11 |
+
pad_token: '[PAD]'
|
| 12 |
+
pad_token_type_id: 0
|