Codeplay: Autotelic Learning through Collaborative Self-Play in Programming Environments - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2023

Codeplay: Autotelic Learning through Collaborative Self-Play in Programming Environments

Abstract

Autotelic learning is the training setup where agents learn by setting their own goals and trying to achieve them. However, creatively generating freeform goals is challenging for autotelic agents. We present Codeplay, an algorithm casting autotelic learning as a game between a Setter agent and a Solver agent, where the Setter generates programming puzzles of appropriate difficulty and novelty for the solver and the Solver learns to achieve them. Early experiments with the Setter demonstrates one can effectively control the tradeoff between difficulty of a puzzle and its novelty by tuning the reward of the Setter, a code language model finetuned with deep reinforcement learning.
Fichier principal
Vignette du fichier
_IMOL_Neurips2023__Codeplay-1.pdf (2.49 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-04374993 , version 1 (05-01-2024)

Identifiers

  • HAL Id : hal-04374993 , version 1

Cite

Laetitia Teodorescu, Cédric Colas, Matthew Bowers, Thomas Carta, Pierre-Yves Oudeyer. Codeplay: Autotelic Learning through Collaborative Self-Play in Programming Environments. IMOL 2023 - Intrinsically Motivated Open-ended Learning workshop at NeurIPS 2023, Dec 2023, New Orleans, United States. ⟨hal-04374993⟩

Collections

INRIA INRIA2
33 View
29 Download

Share

Gmail Facebook X LinkedIn More