Hezar: Upload tokenizer and config
Browse files
preprocessor/tokenizer_config.yaml
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: wordpiece_tokenizer
|
| 2 |
+
config_type: preprocessor
|
| 3 |
+
pretrained_path: hezarai/distilbert-base-fa
|
| 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_to_multiple_of: 0
|
| 11 |
+
pad_token_id: 0
|
| 12 |
+
pad_token: '[PAD]'
|
| 13 |
+
pad_token_type_id: 0
|
| 14 |
+
unk_token: '[UNK]'
|
| 15 |
+
special_tokens:
|
| 16 |
+
- '[UNK]'
|
| 17 |
+
- '[SEP]'
|
| 18 |
+
- '[CLS]'
|
| 19 |
+
- '[PAD]'
|
| 20 |
+
- '[MASK]'
|
| 21 |
+
wordpieces_prefix: '##'
|
| 22 |
+
vocab_size: 30000
|
| 23 |
+
min_frequency: 2
|
| 24 |
+
limit_alphabet: 1000
|
| 25 |
+
initial_alphabet: []
|
| 26 |
+
show_progress: true
|