naps-llama-3.1-8B-instruct-v0.3 — benchmark results

NAPS-ai's Korean LoRA fine-tune of Meta's Llama 3.1 8B Instruct, trained with Unsloth. Provider: Other. Released 2024-09-02. Access: Open.

Unified ELO 1413 ± 34, rank #1168 of 1776 rated models, from 7 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open LLM Leaderboard - MATH Level 519.03Score69.3
Open Korean LLM Leaderboard262.04Average Score (%)64.7
Open LLM Leaderboard - IFEval53.91Score64.7
Open LLM Leaderboard - GPQA6.6Score55.6
Open LLM Leaderboard - MMLU-Pro26.65Score48.8
Open LLM Leaderboard - BBH26.27Score40.8
Open LLM Leaderboard - MuSR7.21Score33.8

Interactive version: theaggregate.ai/model?slug=naps-llama-3-1-8b-instruct-v0-3 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.