Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,29 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# EvoCodeBench
|
2 |
+
|
3 |
+
EvoCodeBench is an evolutionary code generation benchmark aligned with real-world code repositories.
|
4 |
+
|
5 |
+
Details of EvoCodeBench can be found in our paper "EvoCodeBench: An Evolving Code Generation Benchmark Aligned with Real-world Code Repositories".
|
6 |
+
|
7 |
+
**Paper:** https://arxiv.org/pdf/2404.00599.pdf
|
8 |
+
|
9 |
+
**Data and Source Code:** https://github.com/seketeam/EvoCodeBench
|
10 |
+
|
11 |
+
## Features
|
12 |
+
|
13 |
+
Compared to existing benchmarks (e.g., HumanEval), EvoCodeBench has the following features:
|
14 |
+
|
15 |
+
- Aligned with real-world repositories in multiple dimensions, e.g., code distributions and dependency distributions.
|
16 |
+
- Comprehensive annotations (e.g., requirements, code, dependencies, and repos) and robust metrics (Pass@k and Recall@k).
|
17 |
+
- An evolving benchmark to avoid data leakage. The first version - EvoCodeBench-2403 has been released. It contains 275 samples from 25 real-world repositories (created from 10/2023 to 2/2024).
|
18 |
+
|
19 |
+
A sample in EvoCodeBench is shown as follows.
|
20 |
+
|
21 |
+
![Example](images/Example.png)
|
22 |
+
|
23 |
+
## Repository-level Code Generation
|
24 |
+
|
25 |
+
Based on EvoCodeBench, we propose repository-level code generation and evaluate 10 popular LLMs (e.g., gpt-4, gpt-3.5, DeepSeek Coder, StarCoder 2, CodeLLaMa, Gemma, and Qwen 1.5). All prompts and models' completions have been released for further community analysis.
|
26 |
+
|
27 |
+
The evaluation results are shown as follows.
|
28 |
+
|
29 |
+
![Leaderboard](images/LeadBoard.png)
|