WildMBXMarconi-SLERP-7B: benchmark results
jsfs11 release formed by SLERP merging BarryFutureman/WildMarcoroni-Variant1-7B and flemmingmiguel/MBX-7B. The recipe uses distinct attention and MLP interpolation schedules. It uses a 32-layer Mistral decoder configuration and remains distinct from its input checkpoints. Provider: jsfs11. Access: Open.
Aggregate Rating 1466 ± 42, rank #2960 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - ARC Challenge | 73.29 | Normalized accuracy (%) (25-shot) | 96.7 |
| Open LLM Leaderboard v1 - GSM8K | 70.89 | Accuracy (%) (5-shot) | 94.6 |
| Open LLM Leaderboard v1 - HellaSwag | 88.49 | Normalized accuracy (%) (10-shot) | 92.3 |
| Open LLM Leaderboard v1 - WinoGrande | 83.98 | Accuracy (%) (5-shot) | 91.5 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 68.98 | MC2 (%) (0-shot) | 88.3 |
| Open LLM Leaderboard v1 - MMLU | 64.9 | Accuracy (%) (5-shot) | 78.2 |
Interactive version: theaggregate.ai/model?slug=wildmbxmarconi-slerp-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.