{ "vocab_size": 64000, "total_texts": 6179192, "corpus_size_bytes": 17442976221, "training_time_seconds": 1414.3693861961365, "data_sources": { "01_논문_학술": 500042, "02_도서_말뭉치": 500003, "03_전문분야": 509076, "04_웹_뉴스": 501738, "05_법률_행정": 500000, "06_의료_건강": 500000, "07_교육_지식": 214238, "culturax_korean": 200000, "ko_alpaca_instructions": 21149, "korean_textbooks_claude_evol": 200000, "korean_textbooks_code_alpaca": 64006, "korean_textbooks_instructions": 200000, "korean_textbooks_mmlu_all": 97697, "korean_textbooks_tiny": 200000, "korean_textbooks_wikidata": 127470, "wikipedia_korean_2024": 200000, "ko_alpaca_v1.1a": 21149, "ko_orca": 34163, "korean_webtext": 200000, "mc4_korean": 200000, "namuwiki": 200000, "open_ko_instructions": 200000, "cosmopedia_stories_science": 200000, "magpie_reasoning_150k": 148703, "math_instruct": 200000, "python_code_instructions_18k": 18609 }, "date": "2026-08-29 19:30:06", "repo_id": "cooler8/yejin-korean-tokenizer-64k" }