NeuralSynthesis-7B-v0.3: benchmark results
Kukedlc's model-stock merge of six Mistral-7B variants (including YamshadowExperiment28 and M7-7b), built with LazyMergekit. Provider: Kukedlc. Released 2024-04-07. Access: Open.
Aggregate Rating 1467 ± 23, rank #899 of 1656 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 78.13 | MC2 (%) (0-shot) | 98.5 |
| Open LLM Leaderboard v1 - HellaSwag | 89.18 | Normalized accuracy (%) (10-shot) | 98.3 |
| Open LLM Leaderboard v1 - WinoGrande | 84.93 | Accuracy (%) (5-shot) | 96.1 |
| Open LLM Leaderboard v1 - GSM8K | 70.89 | Accuracy (%) (5-shot) | 94.6 |
| Open Chinese LLM - TruthfulQA MC | 63.76 | Accuracy (%) | 94.4 |
| Open LLM Leaderboard v1 - ARC Challenge | 72.7 | Normalized accuracy (%) (25-shot) | 93.1 |
| Open Portuguese LLM - FaQuAD NLI | 77.96 | Macro F1 (%) | 81.9 |
| Open Portuguese LLM - ASSIN2 RTE | 92.24 | Macro F1 (%) | 76.6 |
| Open LLM Leaderboard - MuSR | 43.46 | Score | 74 |
| Open LLM Leaderboard v1 - MMLU | 64.37 | Accuracy (%) (5-shot) | 69 |
| Open Chinese LLM - ARC Challenge | 54.01 | Accuracy (%) | 62 |
| Open Chinese LLM - HellaSwag | 60.88 | Accuracy (%) | 59.5 |
Interactive version: theaggregate.ai/model?slug=neuralsynthesis-7b-v0-3 · How It Works · Data refreshed daily, snapshot 2026-10-20.