Opus-Samantha-Llama-3-8B: benchmark results
A supervised fine-tune made by macadeliccc with AutoSloth, using unsloth/llama-3-8b and the macadeliccc/opus_samantha dataset. This evaluated revision predates the card's later training-source changes. Provider: macadeliccc. Access: Open.
Aggregate Rating 1397 ± 35, rank #6256 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 64.09 | Accuracy (%) (5-shot) | 65.5 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 53.53 | MC2 (%) (0-shot) | 59.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 62.63 | Normalized accuracy (%) (25-shot) | 59 |
| Open LLM Leaderboard v1 - HellaSwag | 82.58 | Normalized accuracy (%) (10-shot) | 52.5 |
| Open LLM Leaderboard v1 - GSM8K | 20.62 | Accuracy (%) (5-shot) | 41.4 |
| Open LLM Leaderboard v1 - WinoGrande | 71.67 | Accuracy (%) (5-shot) | 27.1 |
Interactive version: theaggregate.ai/model?slug=opus-samantha-llama-3-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.