Code-Llama-3-8B: benchmark results
A full Llama-3-8B fine-tune by ajibawa-2023 combining Code-290k-ShareGPT, Code-Feedback, CodeFeedback-Filtered-Instruction and Orca math data, with ChatML prompts. This is distinct from Meta Code Llama releases. Provider: ajibawa-2023. Access: Open.
Aggregate Rating 1292 ± 35, rank #8846 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 49.16 | Accuracy (%) (5-shot) | 28.4 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 44.67 | MC2 (%) (0-shot) | 28 |
| Open LLM Leaderboard v1 - ARC Challenge | 49.74 | Normalized accuracy (%) (25-shot) | 24.2 |
| Open LLM Leaderboard v1 - HellaSwag | 72.77 | Normalized accuracy (%) (10-shot) | 23.2 |
| Open LLM Leaderboard v1 - WinoGrande | 67.56 | Accuracy (%) (5-shot) | 21.6 |
| Open LLM Leaderboard v1 - GSM8K | 2.88 | Accuracy (%) (5-shot) | 20.9 |
Interactive version: theaggregate.ai/model?slug=code-llama-3-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.