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},
}
```
|