Experiment4-7B: benchmark results
A 40-layer Mistral-family checkpoint released as Experiment4-7B. The exact evaluated configuration uses 32,000 vocabulary entries; the release does not document a reproducible training recipe or establish equivalence with other numbered experiments. Provider: Other. Access: Open.
Aggregate Rating 1454 ± 38, rank #3623 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - ARC Challenge | 72.18 | Normalized accuracy (%) (25-shot) | 91.2 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 70.39 | MC2 (%) (0-shot) | 90.3 |
| Open LLM Leaderboard v1 - HellaSwag | 88.09 | Normalized accuracy (%) (10-shot) | 89.5 |
| Open LLM Leaderboard v1 - MMLU | 65.03 | Accuracy (%) (5-shot) | 80.7 |
| Open LLM Leaderboard v1 - WinoGrande | 81.14 | Accuracy (%) (5-shot) | 77.4 |
| Open LLM Leaderboard v1 - GSM8K | 63.46 | Accuracy (%) (5-shot) | 76.8 |
Interactive version: theaggregate.ai/model?slug=experiment4-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.