DeepMind AlphaCode

From GM-RKB
Jump to navigation Jump to search

An DeepMind AlphaCode is a AI-powered coding engine that creates computer programs.



References

2023

  • (AlphaCode Team, 2023) ⇒ AlphaCode Team. (2023). “AlphaCode 2 Technical Report.” Google DeepMind blog.
    • ABSTRACT: AlphaCode (Li et al., 2022) was the first AI system to perform at the level of the median competitor in competitive programming, a difficult reasoning task involving advanced maths, logic and computer science. This paper introduces AlphaCode 2, a new and enhanced system with massively improved performance, powered by Gemini (Gemini Team, Google, 2023). AlphaCode 2 relies on the combination of powerful language models and a bespoke search and reranking mechanism. When evaluated on the same platform as the original AlphaCode, we found that AlphaCode 2 solved 1.7× more problems, and performed better than 85% of competition participants.

2023

  • (Wikipedia, 2023) ⇒ https://en.wikipedia.org/wiki/Google_DeepMind#AlphaCode Retrieved:2023-9-4.
    • In 2022, DeepMind unveiled AlphaCode, an AI-powered coding engine that creates computer programs at a rate comparable to that of an average programmer, with the company testing the system against coding challenges created by Codeforces utilized in human competitive programming competitions.[1] AlphaCode earned a rank equivalent to 54% of the median score on Codeforces after being trained on GitHub data and Codeforce problems and solutions. The program was required to come up with a unique solution and stopped from duplicating answers.

2022

  • https://www.deepmind.com/blog/competitive-programming-with-alphacode
    • QUOTE: ... As part of DeepMind’s mission to solve intelligence, we created a system called AlphaCode that writes computer programs at a competitive level. AlphaCode achieved an estimated rank within the top 54% of participants in programming competitions by solving new problems that require a combination of critical thinking, logic, algorithms, coding, and natural language understanding. ...
    • SUMMARY:
      • Google DeepMind created an AI system called AlphaCode that can generate computer programs to solve competitive programming problems.
      • Competitive programming involves contests where participants have to write code to solve novel problems in a limited time. It tests critical thinking and coding skills.
      • AlphaCode was trained on GitHub code and a dataset of competitive programming problems. It can generate many candidate solutions to a new problem, filter and rank them, and submit the best ones.
      • AlphaCode was evaluated on 10 recent contests from Codeforces. It scored around the level of the median human competitor, placing in the top 54%.
      • This marks the first time an AI code generation system has reached a competitive level in programming contests requiring creative problem solving.
      • The results demonstrate progress in critical thinking abilities for AI. AlphaCode had to invent novel algorithms, not just copy solutions.
      • DeepMind released the competition programming dataset to advance research into problem solving and code generation.
      • While not close to winning contests yet, AlphaCode shows the potential for AI to enhance productivity and open up coding to more people.