Qwen2-7B-ko-Instruct-orpo-ver_2.0_wo_chat Benchmark Results

Qwen2-7B-ko-Instruct-orpo-ver_2.0_wo_chat: benchmark results

Provider: Alibaba. Access: Open.

Unified ELO 1504 ± 17, rank #1073 of 2656 rated models, from 10 benchmark results.

Benchmark	Score	Metric	Percentile
Open Korean LLM - KorNAT-CKA	34.45	Normalized accuracy (%)	93.9
Open Korean LLM - Ko-EQ-Bench	36.33	EQ-Bench score	86.7
Open Korean LLM Leaderboard	42.12	Average Score (%)	84.8
Open Ko-LLM Leaderboard	42.12	Average (%)	84.7
Open Korean LLM - Ko-GSM8K (flexible extract)	42.61	Exact match, flexible extract (%)	77.6
Open Korean LLM - KorNAT-Harmless	63.39	Normalized accuracy (%)	62
Open Korean LLM - Ko-Winogrande	60.38	Accuracy (%)	53.7
Open Korean LLM - Ko-IFEval	28.14	Strict accuracy, prompt/instruction mean (%)	46.5
Open Korean LLM - KorNAT-Helpful	45.8	Normalized accuracy (%)	39.5
Open Korean LLM - Ko-GPQA-Diamond	21.72	Normalized accuracy (%)	29.4