| { | |
| "custom_vocab_files": [ | |
| "/home/usuaris/veu/casimiro.pio.carrino/projects/corpus-utils-lm/corpora/bio/biomedical-clinical.txt" | |
| ], | |
| "vocab_name": "bio-biomedical-clinical-vocab-52k", | |
| "tokenizer": "bbpe", | |
| "lowercase": false, | |
| "vocab_size": 52000, | |
| "min_frequency": 10, | |
| "extra_tokens": [], | |
| "limit_alphabet": 1000, | |
| "no_show_progress": false, | |
| "strip_accents": false, | |
| "no_handle_chinese_chars": false, | |
| "no_clean_text": false, | |
| "reserve_tokens": 0, | |
| "use_tokenizers": false, | |
| "no_fairseq": false, | |
| "files": [ | |
| "/home/usuaris/veu/casimiro.pio.carrino/projects/corpus-utils-lm/corpora/bio/biomedical-clinical.txt" | |
| ], | |
| "output_root_path": "/home/usuaris/veu/casimiro.pio.carrino/projects/corpus-utils-lm/output/model-ready_output/bio-biomedical-clinical-vocab-52k-2021-04-26-0955-3a71-240f", | |
| "commit_hash": "3a7116cf776527c411869becbe6fad8b9e3f5e56" | |
| } |