gemma-7B-openhermes-v0.80: benchmark results
lemon-mint's Gemma 7B OpenHermes v0.80, an instruction-following fine-tune of Google's Gemma 1.1 7B IT on the OpenHermes-2.5 dataset for open-ended conversation. Provider: lemon-mint. Released 2024-03-01. Access: Open.
Aggregate Rating 1374 ± 17, rank #1282 of 1635 rated models, from 26 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Harmless | 64.91 | Normalized accuracy (%) | 77.5 |
| Open Korean LLM - Ko-EQ-Bench | 28.09 | EQ-Bench score | 75.5 |
| Open Korean LLM - KorNAT-Helpful | 48.68 | Normalized accuracy (%) | 71.3 |
| Open Korean LLM - Ko-IFEval | 34.02 | Strict accuracy, prompt/instruction mean (%) | 69.6 |
| Open Korean LLM Leaderboard | 36.73 | Average Score (%) | 58.3 |
| Open LLM Leaderboard v1 - GSM8K | 43.37 | Accuracy (%) (5-shot) | 57.6 |
| Open Korean LLM - KorNAT-CKA | 25.3 | Normalized accuracy (%) | 56.7 |
| Open Korean LLM - Ko-GPQA-Diamond | 23.74 | Normalized accuracy (%) | 55.6 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 17.59 | Exact match, flexible extract (%) | 48.4 |
| Open LLM Leaderboard v1 - MMLU | 56.2 | Accuracy (%) (5-shot) | 38.5 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 47.06 | MC2 (%) (0-shot) | 36.3 |
| Open LLM Leaderboard v1 - ARC Challenge | 54.35 | Normalized accuracy (%) (25-shot) | 31.2 |
Interactive version: theaggregate.ai/model?slug=gemma-7b-openhermes-v0-80 · How It Works · Data refreshed daily, snapshot 2026-11-05.