---
license: mit
task_categories:
- text-to-text
tags:
  - code-translation
  - assembly-language
  - x86
  - arm
  - risc-v
  - cisco-to-risc
  - transpilation
dataset_info:
  features:
  - name: file
    dtype: string
  - name: x86
    dtype: string
  - name: arm
    dtype: string
  splits:
  - name: train
    num_bytes: 3020909
    num_examples: 66
  download_size: 998544
  dataset_size: 3020909
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

This dataset contains pairs of x86 and ARM assembly code, used for training a language model to perform CISC-to-RISC transpilation.

Check out more details here:

- **Paper:** [Guaranteed Guess: A Language Modeling Approach for CISC-to-RISC Transpilation with Testing Guarantees](https://huggingface.co/papers/2506.14606)
- **Code:** [https://github.com/ahmedheakl/Guaranteed-Guess](https://github.com/ahmedheakl/Guaranteed-Guess)
- **Project Page:** [https://ahmedheakl.github.io/Guaranteed-Guess/](https://ahmedheakl.github.io/Guaranteed-Guess/)

The dataset is structured as follows: Each example contains an x86 assembly instruction and its corresponding ARM assembly instruction. (Further details on the data structure can be found in the existing YAML metadata.)