Catégorie
Artificial Intelligence

CodaLab/CodaBench in 2024

CodaLab en 2024

CodaLab/CodaBench in 2024

  • Go beyond
  • News on the same topic

Share

twlkml
Supported by the DATAIA Institute, CodaBench is an open source platform for organizing AI benchmarks. Flexible and powerful, it lets you define tasks and the interface for code (algorithm) submissions. Zoom in.
Corps de texte

What is CodaLab?

CodaLab, the world's first competitive platform for artificial intelligence, catalyzes innovation and scientific research by providing a playground for solving complex societal challenges.

Artificial intelligence (AI) has emerged as one of the drivers of innovation and societal transformation. According to a study published last January, its rise is based in particular on platforms for scientific and technical challenges. These offer fertile ground where researchers and companies can test new ideas to solve complex problems. “By enabling independent and impartial assessments, these platforms increase the credibility and reliability of the advances made,” adds Isabelle Guyon, head of scientific leadership at CodaLab.

CodaLab is a fully open-source platform developed by researchers and engineers at the Laboratoire Interdisciplinaire des Sciences du Numérique (LISN - CNRS/Université Paris-Saclay).


What is CodaBench?

Codabench is an open source platform for organizing AI benchmarks. It's flexible and powerful, yet easy to use. You define the tasks (e.g. datasets and success metrics), then the interface for code submissions (algorithms), add a few pages of documentation, perform an upload and that's it! Your benchmark is created, ready to accept new algorithm submissions. Everything can be fully customized, including the scoring program code. Organizers can even connect their own calculators to their benchmarks, giving them unlimited computing power. Participants can test their methods, get real-time feedback and results on a competitive leaderboard, detailed graphs and more.

Codabench can also be used to create reverse benchmarks. In this case, the roles of data sets and algorithms are reversed. In this scenario, you define the benchmark algorithms and the participants supply the data sets.

Go beyond

Chiffre
10k
Texte
users in October 2024
Chiffre
500
Texte
daily submissions
Chiffre
32k
Texte
new users in 2024
Chiffre
323,000
Texte
submissions
Chiffre
143
Texte
pull requests
News on the same topic
Bandeau image
fleche
Artificial Intelligence
Bandeau image
fleche
Artificial Intelligence