L3-RP_io: benchmark results
Azazelle's TIES merge, whose published recipe uses meta-llama/Meta-Llama-3-8B-Instruct as its base and ResplendentAI/Aura_Uncensored_l3_8B, ResplendentAI/Kei_Llama3_8B, Undi95/Llama-3-Unholy-8B, vicgalle/Roleplay-Llama-3-8B as merge inputs. Its published Llama configuration specifies 32 decoder layers. Provider: Azazelle. Access: Open.
Aggregate Rating 1428 ± 35, rank #4772 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 67.92 | Accuracy (%) (5-shot) | 90.2 |
| Open LLM Leaderboard v1 - GSM8K | 67.85 | Accuracy (%) (5-shot) | 84.5 |
| Open LLM Leaderboard v1 - ARC Challenge | 63.05 | Normalized accuracy (%) (25-shot) | 61 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 52.9 | MC2 (%) (0-shot) | 57.8 |
| Open LLM Leaderboard v1 - WinoGrande | 75.69 | Accuracy (%) (5-shot) | 43.5 |
| Open LLM Leaderboard v1 - HellaSwag | 79.86 | Normalized accuracy (%) (10-shot) | 39.4 |
Interactive version: theaggregate.ai/model?slug=l3-rp-io · How It Works · Data refreshed daily, snapshot 2026-11-08.