Sheared-LLaMA-2.7B: benchmark results
Provider: Princeton NLP. Released 2023-10-10. Access: Open.
Aggregate Rating 1161 ± 26, rank #9826 of 10916 rated models, from 13 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - GPQA | 27.52 | Score | 30.2 |
| PIQA (PapersWithCode) | 75.8 | Accuracy | 30 |
| Open LLM Leaderboard v1 - HellaSwag | 71.01 | Normalized accuracy (%) (10-shot) | 21.1 |
| Open LLM Leaderboard v1 - WinoGrande | 67.01 | Accuracy (%) (5-shot) | 21 |
| Open LLM Leaderboard - IFEval | 24.17 | Score | 18.4 |
| Open LLM Leaderboard v1 - ARC Challenge | 41.72 | Normalized accuracy (%) (25-shot) | 17.9 |
| Open LLM Leaderboard - MuSR | 35.67 | Score | 16.1 |
| Open LLM Leaderboard v1 - GSM8K | 1.06 | Accuracy (%) (5-shot) | 15.4 |
| Open LLM Leaderboard - BBH | 32.59 | Score | 14.3 |
| Open LLM Leaderboard v1 - MMLU | 26.92 | Accuracy (%) (5-shot) | 14.2 |
| Open LLM Leaderboard - MATH Level 5 | 1.28 | Score | 9.2 |
| Open LLM Leaderboard - MMLU-Pro | 11.87 | Score | 9.2 |
Interactive version: theaggregate.ai/model?slug=sheared-llama-2-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.