Delulu - CodeBLEU: leaderboard
Metric: CodeBLEU (0-100): n-gram, syntax-tree and dataflow agreement with the gold completion, on all 1,951 Delulu fill-in-the-middle samples (seven languages, four injected hallucination types), greedy decoding, at most 256 tokens; base FIM models are truncated to the gold completion's line count; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 Coder 32B Instruct | 53.1 |
| 2 | Qwen 2.5 Coder 14B Instruct | 51.5 |
| 3 | CodeLlama-13B-hf | 47.8 |
| 4 | Qwen 2.5 Coder 7B Instruct | 47.2 |
| 5 | Qwen 2.5 Coder 3B Instruct | 47.2 |
| 6 | starcoder2-15B | 46.6 |
| 7 | CodeLlama-7B-hf | 46.3 |
| 8 | starcoder2-7B | 44.5 |
| 9 | Qwen2.5-Coder-1.5B-Instruct | 42.3 |
Interactive version: theaggregate.ai/benchmark?slug=delulu-codebleu · How It Works · Data refreshed daily, snapshot 2026-10-07.