Llama-3-Instruct-8B-SimPO-SPPO-Iter3-merge: benchmark results
grimjim's Llama-3-Instruct-8B-SimPO-SPPO-Iter3-merge, a mergekit merge combining Princeton's Llama-3-Instruct-8B-SimPO and UCLA-AGI's Llama-3-Instruct-8B-SPPO-Iter3 checkpoints. Provider: grimjim. Released 2024-06-28. Access: Open.
Aggregate Rating 1436 ± 17, rank #1065 of 1656 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 51.13 | Normalized accuracy (%) | 89.1 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 54.44 | Exact match, flexible extract (%) | 88.5 |
| Open Korean LLM - Ko-IFEval | 41.55 | Strict accuracy, prompt/instruction mean (%) | 84.8 |
| Open Korean LLM - Ko-EQ-Bench | 32.86 | EQ-Bench score | 82.3 |
| Open LLM Leaderboard - IFEval | 68.06 | Score | 80.7 |
| Open Korean LLM Leaderboard | 40.36 | Average Score (%) | 76.6 |
| Open Ko-LLM Leaderboard | 40.36 | Average (%) | 76.4 |
| Open Korean LLM - Ko-GPQA-Diamond | 25.25 | Normalized accuracy (%) | 75.9 |
| Open Korean LLM - KorNAT-Harmless | 64.43 | Normalized accuracy (%) | 74.2 |
| Open Portuguese LLM - OAB Exams | 49.48 | Accuracy (%) | 72.9 |
| Open Portuguese LLM - ENEM | 68.65 | Accuracy (%) | 70.7 |
| Open Chinese LLM - TruthfulQA MC | 56.24 | Accuracy (%) | 67.2 |
Interactive version: theaggregate.ai/model?slug=llama-3-instruct-8b-simpo-sppo-iter3-merge · How It Works · Data refreshed daily, snapshot 2026-10-20.