Noromaid-7B-v0.2 [float16] [rev: bc35358ec19c]: benchmark results
A Mistral-7B-v0.1 full fine-tune by NeverSleep, credited to Undi and IkariDev. The card reports two epochs and 4,000 steps on no_robots and private Aesir role-play data, with custom or Alpaca prompting. Provider: NeverSleep. Access: Open.
Aggregate Rating 1403 ± 33, rank #5943 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - HellaSwag | 84.97 | Normalized accuracy (%) (10-shot) | 70.5 |
| Open LLM Leaderboard v1 - WinoGrande | 78.37 | Accuracy (%) (5-shot) | 62.7 |
| Open LLM Leaderboard v1 - MMLU | 62.99 | Accuracy (%) (5-shot) | 57.8 |
| Open LLM Leaderboard v1 - ARC Challenge | 62.03 | Normalized accuracy (%) (25-shot) | 55.6 |
| Open LLM Leaderboard v1 - GSM8K | 36.24 | Accuracy (%) (5-shot) | 51.7 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 46.07 | MC2 (%) (0-shot) | 33.3 |
Interactive version: theaggregate.ai/model?slug=noromaid-7b-v0-2-float16-rev-bc35358ec19c · How It Works · Data refreshed daily, snapshot 2026-11-08.