Qwen3-0.6B-JSON-SFT

ํšŒ์˜๋กโ†’JSON ์ถ”์ถœ SFT (ํ•™์ƒ 0.6B ๊ต์ฒด ์‹คํ—˜, ์‹œ๋‚˜๋ฆฌ์˜ค 1)

2026-07-19 sllm_practice ์ปค๋ฆฌํ˜๋Ÿผ ๊ฐœ์„  ์‹คํ—˜ ์‚ฐ์ถœ๋ฌผ (fa2e1ce์˜ experiments/results/REPORT.md ์ฐธ์กฐ). ํ‰๊ฐ€ ์ง€ํ‘œ๋Š” json_eval.py (parse_rate / schema_compliance / semantics_pass_rate / field_f1_mean).

Qwen/Qwen3-0.6B๋ฅผ meeting_to_json_train.jsonl 1,199๊ฐœ๋กœ full FT (no-think). old_eval 300๊ฐœ ๊ธฐ์ค€ parse 98.0 / schema 96.3 / semantics 96.1 / field_F1 51.4% (base 41.6%). ํ•™์Šต 15.4๋ถ„(A40).

Downloads last month
24
Safetensors
Model size
0.6B params
Tensor type
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for NotoriousH2/Qwen3-0.6B-JSON-SFT

Finetuned
Qwen/Qwen3-0.6B
Finetuned
(1078)
this model