Competition-Level Code Generation with AlphaCode (Paper Review)

#ai #alphacode #deepmind

AlphaCode is an automated system that can solve competitive programing exercises. The authors found an interesting combination of language models, large-scale sampling, and clever techniques to filter and subsequently cluster the resulting programs, which lets the system perform on the level of an average competitor in real competitions. In this video, we take a deep dive into AlphaCode's design, architecture, and experimental evaluation. The paper is very well structured and the empirical results are super interesting!

OUTLINE:

0:00 - Intro

2:10 - Paper Overview

3:30 - An example problem from competitive programming

8:00 - AlphaCode system overview

14:00 - Filtering out wrong solutions

17:15 - Clustering equivalent generated programs

21:50 - Model configurations & engineering choices

24:30 - Adding privileged information to the input & more tricks

28:15 - Experimental Results (very interesting!)

Paper: https://storage.googleapis.com/deepmi...

Code: https://github.com/deepmind/code_cont...

Abstract: Programming is a powerful and ubiquitous problem-solving tool. Developing systems that can assist programmers or even generate programs independently could make programming more productive and accessible, yet so far incorporating innovations in AI has proven challenging. Recent large-scale language models have demonstrated an impressive ability to generate code, and are now able to complete simple programming tasks. However, these models still perform poorly when evaluated on more complex, unseen problems that require problem-solving skills beyond simply translating instructions into code. For example, competitive programming problems which require an understanding of algorithms and complex natural language remain extremely challenging. To address this gap, we introduce AlphaCode, a system for code generation that can create novel solutions to these problems that require deeper reasoning. Evaluated on recent programming competitions on the Codeforces platform, AlphaCode achieved on average a ranking of top 54.3% in programming competitions with more than 5,000 participants. We found that three key components were critical to achieve good and reliable performance: (1) an extensive and clean competitive programming dataset for training and evaluation, (2) large and efficient-to-sample transformer-based architectures, and (3) large-scale model sampling to explore the search space, followed by filtering based on program behavior to a small set of submissions.

Authors: Yujia Li, David Choi, Junyoung Chung, Nate Kushman, Julian Schrittwieser, Rémi Leblond, Tom Eccles, James Keeling, Felix Gimeno, Agustin Dal Lago, Thomas Hubert, Peter Choy, Cyprien de Masson d’Autume, Igor Babuschkin, Xinyun Chen, Po-Sen Huang, Johannes Welbl, Sven Gowal, Alexey Cherepanov, James Molloy, Daniel J. Mankowitz, Esme Sutherland Robson, Pushmeet Kohli, Nando de Freitas, Koray Kavukcuoglu and Oriol Vinyals

Links:

Merch: store.ykilcher.com

TabNine Code Completion (Referral): http://bit.ly/tabnine-yannick

YouTube: https://www.youtube.com/c/yannickilcher

Twitter: https://twitter.com/ykilcher

Discord: https://discord.gg/4H8xxDF

BitChute: https://www.bitchute.com/channel/yann...

LinkedIn: https://www.linkedin.com/in/ykilcher

BiliBili: https://space.bilibili.com/2017636191

If you want to support me, the best thing to do is to share out the content :)

If you want to support me financially (completely optional and voluntary, but a lot of people have asked for this):

SubscribeStar: https://www.subscribestar.com/yannick...

Patreon: https://www.patreon.com/yannickilcher

Bitcoin (BTC): bc1q49lsw3q325tr58ygf8sudx2dqfguclvngvy2cq