Llama-3-Lumimaid-8B-v0.1: benchmark results
A Llama 3 role-play release by NeverSleep, credited to Undi and IkariDev. The card describes a mixture of role-play and general instruction datasets and uses the Llama 3 prompting format. Its card also lists the initial Lumimaid fine-tune and the Unholy and LewdPlay derivatives as constituent models. Provider: NeverSleep. Access: Open.
Aggregate Rating 1420 ± 35, rank #5177 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 67.52 | Accuracy (%) (5-shot) | 89.5 |
| Open LLM Leaderboard v1 - GSM8K | 66.19 | Accuracy (%) (5-shot) | 81.3 |
| Open LLM Leaderboard v1 - ARC Challenge | 61.09 | Normalized accuracy (%) (25-shot) | 51.2 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 49.82 | MC2 (%) (0-shot) | 46.1 |
| Open LLM Leaderboard v1 - WinoGrande | 75.06 | Accuracy (%) (5-shot) | 40.1 |
| Open LLM Leaderboard v1 - HellaSwag | 79.63 | Normalized accuracy (%) (10-shot) | 38.5 |
Interactive version: theaggregate.ai/model?slug=llama-3-lumimaid-8b-v0-1 · How It Works · Data refreshed daily, snapshot 2026-11-08.