Llama 3 8B Dpo Sft V1: benchmark results

Provider: Meta.

Aggregate Rating 1436 ± 43, rank #4498 of 10552 rated models, from 9 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open Korean LLM - Ko-GPQA-Diamond26.77Normalized accuracy (%)88.9
Open Korean LLM - Ko-IFEval37.35Strict accuracy, prompt/instruction mean (%)78.1
Open Korean LLM - Ko-GSM8K (flexible extract)41.02Exact match, flexible extract (%)75.4
Open Korean LLM - Ko-EQ-Bench26.74EQ-Bench score73.5
Open Korean LLM Leaderboard39.54Average Score (%)72.7
Open Korean LLM - KorNAT-CKA25.03Normalized accuracy (%)54.7
Open Korean LLM - Ko-Winogrande60.06Accuracy (%)51.1
Open Korean LLM - KorNAT-Helpful45.98Normalized accuracy (%)41
Open Korean LLM - KorNAT-Harmless61.3Normalized accuracy (%)26.9

Interactive version: theaggregate.ai/model?slug=llama-3-8b-dpo-sft-v1 · How It Works · Data refreshed daily, snapshot 2026-11-16.