Starling-LM-7B-beta-ExPO: benchmark results
Provider: UC Berkeley. Released 2024-04-26. Access: Open.
Aggregate Rating 1435 ± 19, rank #4380 of 10916 rated models, from 16 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 65.3 | Accuracy (%) (5-shot) | 83.8 |
| Open LLM Leaderboard v1 - GSM8K | 65.66 | Accuracy (%) (5-shot) | 80.4 |
| Open LLM Leaderboard v1 - WinoGrande | 81.37 | Accuracy (%) (5-shot) | 78.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 67.92 | Normalized accuracy (%) (25-shot) | 78.6 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 57.16 | MC2 (%) (0-shot) | 69.5 |
| Open Chinese LLM - TruthfulQA MC | 55.58 | Accuracy (%) | 63.6 |
| AlpacaEval 2.0 | 26.41 | LC Win Rate (%) | 63.1 |
| Open LLM Leaderboard v1 - HellaSwag | 83.62 | Normalized accuracy (%) (10-shot) | 60.1 |
| Open Chinese LLM - GSM8K | 42.68 | Accuracy (%) | 43.5 |
| WildBench | 31.56 | WB Score Task-Macro | 37.1 |
| Open Chinese LLM - CMMLU | 48.86 | Accuracy (%) | 30.8 |
| Open Chinese LLM Leaderboard | 53.72 | Official Nonzero-Task Mean (%) | 29.6 |
Interactive version: theaggregate.ai/model?slug=starling-lm-7b-beta-expo · How It Works · Data refreshed daily, snapshot 2026-11-08.