Hezar: Upload tokenizer and config
Browse files
preprocessor/tokenizer_config.yaml
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: bpe_tokenizer
|
| 2 |
+
config_type: preprocessor
|
| 3 |
+
max_length: 512
|
| 4 |
+
truncation_strategy: longest_first
|
| 5 |
+
truncation_direction: right
|
| 6 |
+
stride: 0
|
| 7 |
+
padding_strategy: longest
|
| 8 |
+
padding_direction: right
|
| 9 |
+
pad_to_multiple_of: 0
|
| 10 |
+
pad_token_id: 0
|
| 11 |
+
pad_token: <pad>
|
| 12 |
+
pad_token_type_id: 0
|
| 13 |
+
unk_token: <unk>
|
| 14 |
+
special_tokens:
|
| 15 |
+
- <s>
|
| 16 |
+
- <pad>
|
| 17 |
+
- </s>
|
| 18 |
+
- <unk>
|
| 19 |
+
- <mask>
|
| 20 |
+
- <|endoftext|>
|
| 21 |
+
- <|startoftext|>
|
| 22 |
+
- <nl>
|
| 23 |
+
- <hs>
|
| 24 |
+
- <sep>
|
| 25 |
+
- <cls>
|
| 26 |
+
continuing_subword_prefix: ''
|
| 27 |
+
end_of_word_suffix: ''
|
| 28 |
+
fuse_unk: false
|
| 29 |
+
vocab_size: 42000
|
| 30 |
+
min_frequency: 2
|
| 31 |
+
limit_alphabet: 1000
|
| 32 |
+
initial_alphabet: []
|
| 33 |
+
show_progress: true
|