WizardVicuna2-13B-hf: benchmark results
heegyu's WizardVicuna2 13B, Llama 2 13B fine-tuned for 3 epochs on ehartford's unfiltered Wizard-Vicuna 70K dataset. Provider: heegyu. Released 2023-08-07. Access: Open.
Aggregate Rating 1296 ± 14, rank #8800 of 10916 rated models, from 77 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - College Mathematics | 37 | Accuracy (%) | 85.2 |
| MMLU-by-task - High School Mathematics | 29.63 | Accuracy (%) | 72.9 |
| MMLU-by-task - Public Relations | 61.82 | Accuracy (%) | 71.4 |
| MMLU-by-task - Abstract Algebra | 32 | Accuracy (%) | 68.1 |
| MMLU-by-task - Virology | 44.58 | Accuracy (%) | 67.8 |
| MMLU-by-task - Business Ethics | 53 | Accuracy (%) | 66.1 |
| MMLU-by-task - Human Aging | 60.54 | Accuracy (%) | 65.1 |
| MMLU-by-task - International Law | 68.6 | Accuracy (%) | 65.1 |
| MMLU-by-task - Moral Scenarios | 29.05 | Accuracy (%) | 64.9 |
| MMLU-by-task - Professional Accounting | 39.01 | Accuracy (%) | 64.6 |
| MMLU-by-task - High School Microeconomics | 50 | Accuracy (%) | 63.4 |
| MMLU-by-task - College Chemistry | 37 | Accuracy (%) | 62.4 |
Interactive version: theaggregate.ai/model?slug=wizardvicuna2-13b-hf · How It Works · Data refreshed daily, snapshot 2026-11-08.