LogoS-7Bx2-MoE-13B-v0.2: benchmark results
RubielLabarta's SLERP merge of TomGrc's FusionNet 2x7B MoE with its Truthful-DPO variant, an English/Spanish MoE-merging experiment. Provider: RubielLabarta. Released 2024-01-21. Access: Open.
Aggregate Rating 1477 ± 23, rank #861 of 1656 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 88.4 | Accuracy (%) (5-shot) | 100 |
| Open LLM Leaderboard v1 - ARC Challenge | 74.4 | Normalized accuracy (%) (25-shot) | 99.2 |
| Open LLM Leaderboard v1 - HellaSwag | 89.09 | Normalized accuracy (%) (10-shot) | 96.9 |
| Open LLM Leaderboard v1 - GSM8K | 71.57 | Accuracy (%) (5-shot) | 96.1 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 74.53 | MC2 (%) (0-shot) | 95 |
| Open Chinese LLM - TruthfulQA MC | 62.12 | Accuracy (%) | 89.9 |
| Open Chinese LLM - WinoGrande | 67.48 | Accuracy (%) | 83.4 |
| Open Portuguese LLM - FaQuAD NLI | 78.03 | Macro F1 (%) | 82.3 |
| Open Portuguese LLM - ASSIN2 RTE | 92.65 | Macro F1 (%) | 81.7 |
| Open LLM Leaderboard v1 - MMLU | 64.9 | Accuracy (%) (5-shot) | 78.2 |
| Open Chinese LLM - ARC Challenge | 54.1 | Accuracy (%) | 63.3 |
| Open LLM Leaderboard - MuSR | 42.26 | Score | 63 |
Interactive version: theaggregate.ai/model?slug=logos-7bx2-moe-13b-v0-2 · How It Works · Data refreshed daily, snapshot 2026-10-20.