rombos_Replete-Coder-Llama3-8B: benchmark results
rombodawg's Replete-Coder Llama3 8B, trained on 3.9M lines of 75% coding and 25% general instruction data across over 100 languages. Provider: rombodawg. Released 2024-10-06. Access: Open.
Aggregate Rating 1394 ± 17, rank #1246 of 1656 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-EQ-Bench | 27 | EQ-Bench score | 73.9 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 37.83 | Exact match, flexible extract (%) | 72.1 |
| Open Chinese LLM - WinoGrande | 64.33 | Accuracy (%) | 66.1 |
| Open Ko-LLM Leaderboard | 38.29 | Average (%) | 65.9 |
| Open Korean LLM Leaderboard | 38.29 | Average Score (%) | 65.9 |
| Open Korean LLM - Ko-IFEval | 33.13 | Strict accuracy, prompt/instruction mean (%) | 64.9 |
| Open Korean LLM - KorNAT-Helpful | 47.35 | Normalized accuracy (%) | 55.8 |
| Open Portuguese LLM - ENEM | 63.33 | Accuracy (%) | 55.1 |
| Open LLM Leaderboard - IFEval | 47.14 | Score | 54.1 |
| Open Portuguese LLM - OAB Exams | 42.41 | Accuracy (%) | 51.7 |
| Open Portuguese LLM - BLUEX | 51.18 | Accuracy (%) | 48.9 |
| Open Korean LLM - Ko-Winogrande | 59.67 | Accuracy (%) | 48.1 |
Interactive version: theaggregate.ai/model?slug=rombos-replete-coder-llama3-8b · How It Works · Data refreshed daily, snapshot 2026-10-20.