Update README.md
Browse files
README.md
CHANGED
@@ -1,4 +1,31 @@
|
|
1 |
---
|
2 |
license: mit
|
3 |
---
|
4 |
-
This is the evaluation benchmark of [LLM4Decompile](https://github.com/albertan017/LLM4Decompile) project.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: mit
|
3 |
---
|
4 |
+
This is the evaluation benchmark of [LLM4Decompile](https://github.com/albertan017/LLM4Decompile) project.
|
5 |
+
|
6 |
+
## About
|
7 |
+
* **Decompile-Bench-Eval** includes manually crafted binaries from the well-established HumanEval and MBPP, alongside the compiled GitHub repositories released after 2025 to mitigate data leakage issues.
|
8 |
+
* **C and C++ Support:** These datasets include both **C** and **C++** source code, whereas earlier models (LLM4Decompile-V1.5, V2) and the HumanEval-Decompile dataset were limited to C only.
|
9 |
+
|
10 |
+
## Columns
|
11 |
+
It contains three splits, huameval, mbpp, and github2025. We also provide a [json verison](https://github.com/albertan017/LLM4Decompile/tree/main/decompile-bench/data) for the data.
|
12 |
+
They contains the following columns:
|
13 |
+
```
|
14 |
+
{
|
15 |
+
"index":"index of the function",
|
16 |
+
"func_name":"demangled name for he function",
|
17 |
+
"func_dep":"function dependecies (includes, help functions), or the path to the source code",
|
18 |
+
"func":"source code",
|
19 |
+
"test":"unit tests for the function, empty for github data",
|
20 |
+
"opt":"optimization, O0, O1, O2, O3",
|
21 |
+
"language":"language, c or cpp",
|
22 |
+
"asm":"assembly",
|
23 |
+
"ida_asm":"assembly from ida pro",
|
24 |
+
"ida_pseudo":"decompiled results (pseudo code) from ida pro",
|
25 |
+
"ghidra_asm":"assembly from ghidra",
|
26 |
+
"ghidra_pseudo":"decompiled results (pseudo code) from ghidra"
|
27 |
+
}
|
28 |
+
```
|
29 |
+
|
30 |
+
## Others
|
31 |
+
For more details, please check [LLM4Decompile](https://github.com/albertan017/LLM4Decompile) project.
|