K2S3-Mistral-7B-v1.1: benchmark results
K2S3 release v1.1 published by Changgil: a full-parameter supervised fine-tune of mistralai/Mistral-7B-v0.1, using alpaca-gpt4-data and OpenOrca samples. The release configuration uses 32 Mistral decoder layers; versioned siblings are not interchangeable checkpoints. Provider: Changgil. Access: Open.
Aggregate Rating 1388 ± 33, rank #6622 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - GSM8K | 53.53 | Accuracy (%) (5-shot) | 65.2 |
| Open LLM Leaderboard v1 - MMLU | 60.56 | Accuracy (%) (5-shot) | 48.8 |
| Open LLM Leaderboard v1 - WinoGrande | 76.48 | Accuracy (%) (5-shot) | 48.5 |
| Open LLM Leaderboard v1 - ARC Challenge | 59.47 | Normalized accuracy (%) (25-shot) | 43.5 |
| Open LLM Leaderboard v1 - HellaSwag | 80.75 | Normalized accuracy (%) (10-shot) | 42.2 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 45.54 | MC2 (%) (0-shot) | 31.3 |
Interactive version: theaggregate.ai/model?slug=k2s3-mistral-7b-v1-1 · How It Works · Data refreshed daily, snapshot 2026-11-08.