Smaug-Llama-3-70B-Instruct-32K: benchmark results
Abacus.AI's Smaug fine-tune of Llama 3 70B Instruct with context extended to 32K via PoSE training. Provider: Abacus.AI. Released 2024-06-11. Access: Open.
Aggregate Rating 1576 ± 45, rank #1307 of 10916 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - IFEval | 77.61 | Score | 93.5 |
| Open Chinese LLM - CMMLU | 71.66 | Accuracy (%) | 93.2 |
| Open Chinese LLM - GSM8K | 68.08 | Accuracy (%) | 92.6 |
| Open LLM Leaderboard - BBH | 64.93 | Score | 90.9 |
| Open LLM Leaderboard - MMLU-Pro | 47.65 | Score | 86.4 |
| Open Chinese LLM Leaderboard | 67.04 | Official Nonzero-Task Mean (%) | 85.8 |
| Open LLM Leaderboard - MATH Level 5 | 27.49 | Score | 79.7 |
| Open Chinese LLM - C-Eval Semantic | 86.22 | Accuracy (%) | 79 |
| Open Chinese LLM - ARC Challenge | 57.94 | Accuracy (%) | 78 |
| Open Chinese LLM - HellaSwag | 63.79 | Accuracy (%) | 74.6 |
| Open LLM Leaderboard - MuSR | 42.08 | Score | 61.6 |
| Open LLM Leaderboard - GPQA | 29.61 | Score | 51.6 |
Interactive version: theaggregate.ai/model?slug=smaug-llama-3-70b-instruct-32k · How It Works · Data refreshed daily, snapshot 2026-11-08.