OpenHermes-2.5-Code-290k-13B: benchmark results
A full Llama-2 fine-tune by ajibawa-2023 on approximately 1.29 million conversation sets combining the OpenHermes-2.5 and Code-290k-ShareGPT datasets; OpenHermes names the training data rather than a merged parent checkpoint. Provider: ajibawa-2023. Access: Open.
Aggregate Rating 1367 ± 32, rank #7233 of 10916 rated models, from 7 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - GSM8K | 58.3 | Accuracy (%) (5-shot) | 70 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 52.5 | MC2 (%) (0-shot) | 56.3 |
| EvalPlus (HumanEval+ & MBPP+) | 47.3 | Pass@1 avg (%) | 41.5 |
| Open LLM Leaderboard v1 - HellaSwag | 80.48 | Normalized accuracy (%) (10-shot) | 41.1 |
| Open LLM Leaderboard v1 - MMLU | 56.53 | Accuracy (%) (5-shot) | 39.4 |
| Open LLM Leaderboard v1 - WinoGrande | 74.82 | Accuracy (%) (5-shot) | 38.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 57.34 | Normalized accuracy (%) (25-shot) | 37.3 |
Interactive version: theaggregate.ai/model?slug=openhermes-2-5-code-290k-13b · How It Works · Data refreshed daily, snapshot 2026-11-08.