Llama 3 8B Base Coder V3.5 10k: benchmark results
A Llama-3 base-model coding fine-tune by rombodawg on Replete-AI/code_test_dataset_10k, trained for five epochs. Provider: rombodawg. Access: Open.
Aggregate Rating 1402 ± 34, rank #5960 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 65.38 | Accuracy (%) (5-shot) | 84.2 |
| Open LLM Leaderboard v1 - WinoGrande | 78.3 | Accuracy (%) (5-shot) | 62 |
| Open LLM Leaderboard v1 - GSM8K | 42 | Accuracy (%) (5-shot) | 56.7 |
| Open LLM Leaderboard v1 - HellaSwag | 81.13 | Normalized accuracy (%) (10-shot) | 43.7 |
| Open LLM Leaderboard v1 - ARC Challenge | 57.25 | Normalized accuracy (%) (25-shot) | 37 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 45.13 | MC2 (%) (0-shot) | 29.8 |
Interactive version: theaggregate.ai/model?slug=llama-3-8b-base-coder-v3-5-10k · How It Works · Data refreshed daily, snapshot 2026-11-08.