File size: 3,628 Bytes
cc96c6f
 
 
dcc4c1a
cc96c6f
 
 
 
 
 
 
01d4118
c984dac
 
cc96c6f
 
 
 
 
 
 
d8a1762
cc96c6f
6a16ffd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a405510
6a16ffd
 
 
 
a405510
 
 
 
 
 
6a16ffd
 
 
 
 
 
 
 
c984dac
6a16ffd
c984dac
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
---
library_name: peft
license: mit
base_model: LLM4Binary/llm4decompile-6.7b-v1.5
tags:
- llama-factory
- lora
- generated_from_trainer
model-index:
- name: ref-decompile
  results: []
pipeline_tag: text-generation
datasets:
- ylfeng/ReF-Decompile-dataset
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# ReF Decompile

This model is a fine-tuned version of [LLM4Binary/llm4decompile-6.7b-v1.5](https://huggingface.co/LLM4Binary/llm4decompile-6.7b-v1.5).

## Results

<table class="table">
    <tr class="header">
        <th rowspan="2">Model/Metrics</th>
        <th colspan="5">Re-executability Rate (%)</th>
        <th colspan="5">Readability (#)</th>
    </tr>
    <tr class="header">
        <td>O0</td><td>O1</td><td>O2</td><td>O3</td><td>AVG</td>
        <td>O0</td><td>O1</td><td>O2</td><td>O3</td><td>AVG</td>
    </tr>
    <tr style="text-align: center;"><td colspan="11"><strong>Rule Based Decompiler</strong></td></tr>
    <tr>
        <td>ghidra</td>
        <td>34.76</td><td>16.46</td><td>15.24</td><td>14.02</td><td>20.12</td>
        <td>2.98</td><td>2.41</td><td>2.52</td><td>2.38</td><td>2.57</td>
    </tr>
    <tr style="text-align: center;"><td colspan="11"><strong>Refine-Based Method</strong></td></tr>
    <tr>
        <td>GPT-4o</td>
        <td>46.95</td><td>34.15</td><td>28.66</td><td>31.10</td><td>35.22</td>
        <td>2.82</td><td>2.35</td><td>2.29</td><td>2.31</td><td>2.44</td>
    </tr>
    <tr>
        <td><a href="https://huggingface.co/LLM4Binary/llm4decompile-6.7b-v2">LLM4Decompile-Ref</a></td>
        <td>74.39</td><td>46.95</td><td>47.56</td><td>42.07</td><td>52.74</td>
        <td>4.08</td><td>3.38</td><td>3.34</td><td>3.19</td><td>3.50</td>
    </tr>
    <tr style="text-align: center;"><td colspan="11"><strong>End-to-End Method</strong></td></tr>
    <tr>
        <td><a href="https://huggingface.co/LLM4Binary/llm4decompile-6.7b-v1.5">LLM4Decompile-End</a></td>
        <td>69.51</td><td>44.51</td><td>39.63</td><td>38.41</td><td>48.02</td>
        <td>4.07</td><td>3.46</td><td>3.40</td><td>3.23</td><td>3.54</td>
    </tr>
    <tr>
        <td><a href="https://github.com/AlongWY/sccdec">FAE Decompile</a></td>
        <td>67.68</td><td>48.78</td><td>45.73</td><td>42.07</td><td>51.07</td>
        <td>3.94</td><td>3.46</td><td>3.40</td><td>3.25</td><td>3.51</td>
    </tr>
    <tr>
        <td><a href="https://github.com/AlongWY/sccdec">FAE Decompile+SCC</a></td>
        <td>70.24</td><td>48.54</td><td>47.56</td><td>43.29</td><td>52.41</td>
        <td>3.97</td><td>3.48</td><td>3.41</td><td>3.23</td><td>3.52</td>
    </tr>
    <tr>
        <td><a href="https://github.com/AlongWY/ReF-Dec">ReF Decompile</a></td>
        <td>85.37</td><td>56.10</td><td>51.83</td><td>52.43</td><td>61.43</td>
        <td>4.13</td><td>3.60</td><td>3.54</td><td>3.49</td><td>3.69</td>
    </tr>
</table>

## Resources

+ [Code](https://github.com/AlongWY/ReF-Dec)
+ [Paper](https://arxiv.org/abs/2502.12221)
+ [Model](https://huggingface.co/ylfeng/ReF-Decompile-lora)
+ [Dataset](https://huggingface.co/datasets/ylfeng/ReF-Decompile-dataset)

## Reference

```bibtex
@misc{feng2025refdecompilerelabelingfunction,
      title={ReF Decompile: Relabeling and Function Call Enhanced Decompile}, 
      author={Yunlong Feng and Bohan Li and Xiaoming Shi and Qingfu Zhu and Wanxiang Che},
      year={2025},
      eprint={2502.12221},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      url={https://arxiv.org/abs/2502.12221}, 
}
```