NeuralTrix-7B-dpo-relaser: benchmark results
Kquant03's separately published NeuralTrix variant. Its configuration points to CultriX/NeuralTrix-7B-dpo, and its inherited card describes a DARE-TIES merge followed by DPO on jondurbin/truthy-dpo-v0.1. The card does not document the exact laser or relaser operation; those release labels remain distinct without a claim of equivalent weights. Provider: Kquant03. Access: Open.
Aggregate Rating 1457 ± 41, rank #3457 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 77.98 | MC2 (%) (0-shot) | 97.9 |
| Open LLM Leaderboard v1 - WinoGrande | 84.06 | Accuracy (%) (5-shot) | 91.8 |
| Open LLM Leaderboard v1 - HellaSwag | 88.41 | Normalized accuracy (%) (10-shot) | 91.6 |
| Open LLM Leaderboard v1 - ARC Challenge | 71.33 | Normalized accuracy (%) (25-shot) | 89 |
| Open LLM Leaderboard v1 - GSM8K | 68.16 | Accuracy (%) (5-shot) | 85.2 |
| Open LLM Leaderboard v1 - MMLU | 64.01 | Accuracy (%) (5-shot) | 64.8 |
Interactive version: theaggregate.ai/model?slug=neuraltrix-7b-dpo-relaser · How It Works · Data refreshed daily, snapshot 2026-11-08.