| { | |
| "corpus_path": "combined_thai_corpus.txt", | |
| "vocab_size": 35590, | |
| "model_type": "unigram", | |
| "min_frequency": 2, | |
| "max_token_length": 16, | |
| "use_thai_pretokenizer": false, | |
| "thai_engine": null, | |
| "normalize_text": false, | |
| "enable_byte_fallback": false, | |
| "dropout": null, | |
| "special_tokens": ["<unk>"], | |
| "training_time": "2025-07-04 00:00:00" | |
| } | |