preprocessor/tokenizer_config.yaml

name: bpe_tokenizer
config_type: preprocessor
max_length: 512
truncation_strategy: longest_first
truncation_direction: right
stride: 0
padding_strategy: longest
padding_direction: right
pad_to_multiple_of: 0
pad_token_type_id: 0
bos_token: <s>
eos_token: </s>
unk_token: <unk>
sep_token: <sep>
pad_token: </s>
cls_token: <cls>
mask_token: <mask>
special_tokens:
- <s>
- <pad>
- </s>
- <unk>
- <mask>
- <|endoftext|>
- <|startoftext|>
- <nl>
- <hs>
- <sep>
- <cls>
continuing_subword_prefix: ''
end_of_word_suffix: ''
fuse_unk: false
vocab_size: 42000
min_frequency: 2
limit_alphabet: 1000
initial_alphabet: []
show_progress: true