GRM-llama3-8B-sftreg: benchmark results
Provider: Other. Access: Open.
Unified ELO 1612 ± 43, rank #417 of 1605 rated models, from 17 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| RewardBench Chat | 98.6 | Accuracy (%) | 98.3 |
| RewardBench Prior Sets (0.5 weight) | 73.09 | Score (%) | 91.3 |
| RewardBench | 85.42 | Score (%) | 84.9 |
| RewardBench Safety | 89.19 | Accuracy (%) | 80.2 |
| RewardBench Reasoning | 92.29 | Accuracy (%) | 78.1 |
| RewardBench Precise IF | 38.75 | Score (%) | 61.7 |
| RewardBench Chat Hard | 67.76 | Accuracy (%) | 61.4 |
| Themis-CodeRewardBench - Security Hardness | 67 | Preference accuracy (%; share of the benchmark's preference | 59.2 |
| RewardBench Ties | 59.81 | Score (%) | 57.4 |
| Themis-CodeRewardBench - Readability and Maintainability | 66.11 | Preference accuracy (%; share of the benchmark's preference | 50 |
| Themis-CodeRewardBench | 71.81 | Preference accuracy (%; share of the benchmark's preference | 49 |
| RewardBench Focus | 68.28 | Score (%) | 40.3 |
Interactive version: theaggregate.ai/model?slug=grm-llama3-8b-sftreg · How It Works · Data refreshed daily, snapshot 2026-09-26.