crammed-bert-legacy / tokenizer_config.json
JonasGeiping's picture
Upload tokenizer
c27e877
raw
history blame contribute delete
534 Bytes
{
"bos_token": "<s>",
"cls_token": "<cls>",
"eos_token": "</s>",
"mask_token": "<mask>",
"model_max_length": 128,
"name_or_path": "/home/jonas/Documents/cmlscratch_backups/cramming/A6000amp_b4096_c5_o3_final/checkpoints/ScriptableMaskedLM_2022-12-21_2.0118",
"pad_token": "<pad>",
"sep_token": "<sep>",
"special_tokens_map_file": "/scratch1/jonas0/c4-subset_WordPiecex32768_e0501aeb87699de7500dc91a54939f44/tokenizer/special_tokens_map.json",
"tokenizer_class": "PreTrainedTokenizerFast",
"unk_token": "<unk>"
}