| # Process config example for dataset | |
| # global parameters | |
| project_name: 'test-process' | |
| dataset_path: '/workspace/thviet/LLMs/Monolingual/toxic/toxic_dedup/c4_part1_nontoxic_dedup.json' | |
| np: 10 # number of subprocess to process your dataset | |
| text_keys: 'text' | |
| export_path: '/workspace/thviet/LLMs/Monolingual/toxic/toxic_dedup/c4_part1_nontoxic_dedup.jsonl' | |
| #use_cache: true | |
| #cache_compress: 'gzip' | |
| #suffixes: ['.json'] | |
| #use_checkpoint: true | |
| #executor_type: default # type of executor, support "default" or "ray" for now. | |
| # process schedule | |
| # a list of several process operators with their arguments | |
| process: | |
| - fix_unicode_mapper: | |
| - sentence_split_mapper: # split text to multiple sentences and join them with '\n' | |
| lang: vi | |
| - language_id_score_filter: # filter text in specific language with language scores larger than a specific max value | |
| lang: vi # keep text in what language | |
| min_score: 0.8 |