Overview

The WUKONG PROJECT is a Large Language Model (LLM) initiative focused on mathematics, code evaluation, and content translation. The project aims to enhance the multilingual assessment of AI models in mathematical reasoning and programming capabilities by creating diverse benchmark datasets.

Objectives

The primary goals of the WUKONG PROJECT include:

• Producing high-quality samples for mathematics and code evaluation.
  
• Translating existing mathematical and coding benchmarks into multiple languages.
  
• Constructing an Olympiad-level multilingual evaluation benchmark for strong inference models such as O1/R1.
  
• Ensuring daily assessments of model capabilities using translated benchmarks.
  

Project Requests

Request 1: Mathematics and Code Evaluation Sample Production

• Translate mathematical and coding benchmarks into multiple languages.
  
• Incorporate these benchmarks into the daily assessment of LLM capabilities.
  
• Ensure accuracy and consistency in translation by utilizing translators proficient in LaTeX grammar for mathematical notation.
  

Dragon_Math_Latex_Translation
Task Sample 1

Request 2: Multilingual Mathematics Challenge Benchmark

• Develop an Olympiad-level multilingual evaluation benchmark tailored for advanced inference models.
  
• Translate mathematical test questions into several languages to support diverse language models.
  

Legend_Math_MTPE_Instruction
Legend_Math Translation_Instruction
Task Sample 2

Work Scope

The WUKONG PROJECT involves the following key tasks:

• Term Search & Verification – Identifying and verifying mathematical and programming terminology across multiple languages.
  
• Content Translation – Translating mathematical and coding content while maintaining accuracy and contextual integrity.
  

Resource Requirements

Skilled translators with expertise in LaTeX grammar.
Proficiency in mathematical terminology and coding languages.
Knowledge of benchmarking methodologies for AI assessment.

Conclusion

The WUKONG PROJECT is committed to advancing the multilingual evaluation of AI models in mathematics and programming. By developing comprehensive benchmarks and ensuring accurate translations, the project aims to contribute to the improvement of AI inference and problem-solving capabilities on a global scale.

Payment information for translators