Scarlett-Llama-3-8B-v1.0: benchmark results
A full Llama-3-8B fine-tune by ajibawa-2023 on over 10,000 conversation sets covering philosophy, advice, jokes and coding. The publisher explicitly identifies v1.0 as an updated release of Scarlett-Llama-3-8B; both checkpoints remain distinct. Provider: ajibawa-2023. Access: Open.
Aggregate Rating 1420 ± 33, rank #5176 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 66.36 | Accuracy (%) (5-shot) | 87 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 55.98 | MC2 (%) (0-shot) | 66.8 |
| Open LLM Leaderboard v1 - HellaSwag | 83.98 | Normalized accuracy (%) (10-shot) | 62.9 |
| Open LLM Leaderboard v1 - WinoGrande | 77.9 | Accuracy (%) (5-shot) | 59.3 |
| Open LLM Leaderboard v1 - GSM8K | 43.21 | Accuracy (%) (5-shot) | 57.4 |
| Open LLM Leaderboard v1 - ARC Challenge | 62.12 | Normalized accuracy (%) (25-shot) | 56.1 |
Interactive version: theaggregate.ai/model?slug=scarlett-llama-3-8b-v1-0 · How It Works · Data refreshed daily, snapshot 2026-11-08.