multimaster-7B-v6: benchmark results
ibivibiv's Mixtral-style multi-discipline MoE built from LoRA-tuned OpenChat-3.5 7B experts, part of a series testing MoE gate tuning. Provider: ibivibiv. Released 2024-02-24. Access: Open.
Aggregate Rating 1471 ± 23, rank #885 of 1656 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 86.42 | Accuracy (%) (5-shot) | 99.2 |
| Open LLM Leaderboard v1 - HellaSwag | 88.77 | Normalized accuracy (%) (10-shot) | 94 |
| Open LLM Leaderboard v1 - ARC Challenge | 72.78 | Normalized accuracy (%) (25-shot) | 93.5 |
| Open LLM Leaderboard v1 - GSM8K | 70.36 | Accuracy (%) (5-shot) | 92.7 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 70.89 | MC2 (%) (0-shot) | 91 |
| Open Chinese LLM - TruthfulQA MC | 60.33 | Accuracy (%) | 83.4 |
| Open LLM Leaderboard - MuSR | 43.96 | Score | 77.7 |
| Open Portuguese LLM - ASSIN2 RTE | 92.2 | Macro F1 (%) | 76 |
| Open LLM Leaderboard v1 - MMLU | 64.74 | Accuracy (%) (5-shot) | 75.8 |
| Open Chinese LLM - WinoGrande | 65.04 | Accuracy (%) | 70.7 |
| Open Portuguese LLM - FaQuAD NLI | 71.86 | Macro F1 (%) | 64.7 |
| Open Portuguese LLM - BLUEX | 54.8 | Accuracy (%) | 61.4 |
Interactive version: theaggregate.ai/model?slug=multimaster-7b-v6 · How It Works · Data refreshed daily, snapshot 2026-10-20.