K2S3-SOLAR-11B-v2.0: benchmark results
K2S3 release published by Changgil: a full-parameter supervised fine-tune of upstage/SOLAR-10.7B-v1.0 using Korean dictionary, KULLM, thesis-abstract and AI Hub data. The card repeats a v1.0 model-number heading; repository version and explicit base-model evidence are retained without merging the releases. Provider: Changgil. Access: Open.
Aggregate Rating 1220 ± 34, rank #9479 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 47.72 | MC2 (%) (0-shot) | 38.5 |
| Open LLM Leaderboard v1 - MMLU | 37.35 | Accuracy (%) (5-shot) | 19.1 |
| Open LLM Leaderboard v1 - ARC Challenge | 41.64 | Normalized accuracy (%) (25-shot) | 17.8 |
| Open LLM Leaderboard v1 - GSM8K | 1.29 | Accuracy (%) (5-shot) | 16.3 |
| Open LLM Leaderboard v1 - WinoGrande | 63.46 | Accuracy (%) (5-shot) | 16.2 |
| Open LLM Leaderboard v1 - HellaSwag | 61.67 | Normalized accuracy (%) (10-shot) | 15.6 |
Interactive version: theaggregate.ai/model?slug=k2s3-solar-11b-v2-0 · How It Works · Data refreshed daily, snapshot 2026-11-08.