ChimeraLlama-3-8B-v2: benchmark results
mlabonne's ChimeraLlama 3 8B v2, a LazyMergekit merge of six Llama 3 8B fine-tunes including Dolphin, OrpoLlama and Configurable-Llama-3. Provider: mlabonne. Released 2024-04-22. Access: Open.
Aggregate Rating 1421 ± 21, rank #1141 of 1656 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - GSM8K | 73.01 | Accuracy (%) (5-shot) | 98.4 |
| Open LLM Leaderboard v1 - MMLU | 67.83 | Accuracy (%) (5-shot) | 89.9 |
| Open Chinese LLM - GSM8K | 50.72 | Accuracy (%) | 69.4 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 56.53 | MC2 (%) (0-shot) | 68.3 |
| Open LLM Leaderboard v1 - ARC Challenge | 63.82 | Normalized accuracy (%) (25-shot) | 64.4 |
| Open Chinese LLM - TruthfulQA MC | 54.76 | Accuracy (%) | 58 |
| Open Chinese LLM - C-Eval Semantic | 71.46 | Accuracy (%) | 57.4 |
| Open Chinese LLM - CMMLU | 54.16 | Accuracy (%) | 53.7 |
| Open LLM Leaderboard - MMLU-Pro | 35.69 | Score | 53.7 |
| Open LLM Leaderboard - BBH | 50.46 | Score | 50.1 |
| Open LLM Leaderboard - IFEval | 44.69 | Score | 48.8 |
| Open Chinese LLM Leaderboard | 56.89 | Official Nonzero-Task Mean (%) | 48.5 |
Interactive version: theaggregate.ai/model?slug=chimerallama-3-8b-v2 · How It Works · Data refreshed daily, snapshot 2026-10-20.