Matter-0.2-7B-DPO: benchmark results
0-hero's Matter 7B v0.2 DPO, fine-tuned on their Matter dataset curated from over 35 sources totaling 6B tokens, with function calling. Provider: 0-hero. Released 2024-04-13. Access: Open.
Aggregate Rating 1388 ± 28, rank #1261 of 1656 rated models, from 18 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 80.27 | Accuracy (%) (5-shot) | 73.2 |
| Open LLM Leaderboard v1 - GSM8K | 56.94 | Accuracy (%) (5-shot) | 68.8 |
| Open LLM Leaderboard v1 - ARC Challenge | 63.31 | Normalized accuracy (%) (25-shot) | 62.2 |
| Open LLM Leaderboard v1 - MMLU | 62.9 | Accuracy (%) (5-shot) | 57.4 |
| Open LLM Leaderboard v1 - HellaSwag | 83.16 | Normalized accuracy (%) (10-shot) | 56.5 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 50.3 | MC2 (%) (0-shot) | 47.9 |
| Open CoT - LSAT Reading Comprehension | 11.15 | CoT Gain (%) | 46.9 |
| Open CoT - LogiQA | 3.35 | CoT Gain (%) | 39.7 |
| Open LLM Leaderboard - IFEval | 33.03 | Score | 30.9 |
| Open CoT - LSAT Logical Reasoning | 6.67 | CoT Gain (%) | 30.2 |
| Open LLM Leaderboard - MuSR | 38.14 | Score | 30.1 |
| Open CoT Leaderboard | 4.7 | Average CoT Gain (%) | 29 |
Interactive version: theaggregate.ai/model?slug=matter-0-2-7b-dpo · How It Works · Data refreshed daily, snapshot 2026-10-20.