Update preprocessor/tokenizer_config.yaml
Browse files
preprocessor/tokenizer_config.yaml
CHANGED
@@ -1,11 +1,11 @@
|
|
1 |
name: sentencepiece_unigram_tokenizer
|
2 |
config_type: preprocessor
|
3 |
max_length: 512
|
4 |
-
|
5 |
-
|
6 |
stride: 0
|
7 |
-
|
8 |
-
|
9 |
pad_to_multiple_of: 0
|
10 |
pad_token_type_id: 0
|
11 |
bos_token: <s>
|
|
|
1 |
name: sentencepiece_unigram_tokenizer
|
2 |
config_type: preprocessor
|
3 |
max_length: 512
|
4 |
+
truncation: longest_first
|
5 |
+
truncation_side: right
|
6 |
stride: 0
|
7 |
+
padding: longest
|
8 |
+
padding_side: right
|
9 |
pad_to_multiple_of: 0
|
10 |
pad_token_type_id: 0
|
11 |
bos_token: <s>
|