EvoCraft: A New Challenge for Open-Endedness

Grbic, Djordje; Palm, Rasmus; Najarro, Elias; Glanois, Claire; Risi, Sebastian

doi:10.1007/978-3-030-72699-7_21

Cited by 15 publications

(12 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Given the primacy of search with ML, research within open-endedness naturally has focused on refining algorithms for open-ended search, such as those driven by novelty [34,35] or curiosity [36,37]. While such focus has indeed lead to algorithmic progress, there is a growing awareness of the criticality of the environment in which open-ended algorithms are applied [38][39][40][41].…”

Section: Open-endednessmentioning

confidence: 99%

“…That is, the environment limits what can arise within the system and for how long its products can remain interesting. As a result, some have argued for more complex environments for open-endedness, such as video games [38,39], and others have argued that features of the environment should co-evolve with agents [40,42]. Yet a theory for what specific forms of additional such complexity is needed for enduring open-endedness has been lacking.…”

Section: Open-endednessmentioning

confidence: 99%

See 1 more Smart Citation

Evolution through Large Models

Lehman¹,

Gordon²,

Jain³

et al. 2022

Preprint

View full text Add to dashboard Cite

This paper pursues the insight that large language models (LLMs) trained to generate code can vastly improve the effectiveness of mutation operators applied to programs in genetic programming (GP). Because such LLMs benefit from training data that includes sequential changes and modifications, they can approximate likely changes that humans would make. To highlight the breadth of implications of such evolution through large models (ELM), in the main experiment ELM combined with MAP-Elites generates hundreds of thousands of functional examples of Python programs that output working ambulating robots in the Sodarace domain, which the original LLM had never seen in pre-training. These examples then help to bootstrap training a new conditional language model that can output the right walker for a particular terrain. The ability to bootstrap new models that can output appropriate artifacts for a given context in a domain where zero training data was previously available carries implications for open-endedness, deep learning, and reinforcement learning. These implications are explored here in depth in the hope of inspiring new directions of research now opened up by ELM.

show abstract

Section: Open-endednessmentioning

confidence: 99%

Section: Open-endednessmentioning

confidence: 99%

Evolution through Large Models

Lehman¹,

Gordon²,

Jain³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Several simplified Minecraft-inspired simulators were proposed [19], [20] to study the creative space of 3D structures. Grbic et al [21] introduce the problem of open-ended procedural content generation in Minecraft. Sudhakaran et al [22] use the Neural Cellular Automata architecture to produce a fixed structure in Minecraft given a seed or partial structure.…”

Section: Related Workmentioning

confidence: 99%

World-GAN: a Generative Model for Minecraft Worlds

Awiszus

Schubert

Rosenhahn

2021

2021 IEEE Conference on Games (CoG)

View full text Add to dashboard Cite

This work introduces World-GAN, the first method to perform data-driven Procedural Content Generation via Machine Learning in Minecraft from a single example. Based on a 3D Generative Adversarial Network (GAN) architecture, we are able to create arbitrarily sized world snippets from a given sample. We evaluate our approach on creations from the community as well as structures generated with the Minecraft World Generator. Our method is motivated by the dense representations used in Natural Language Processing (NLP) introduced with word2vec [1]. The proposed block2vec representations make World-GAN independent from the number of different blocks, which can vary a lot in Minecraft, and enable the generation of larger levels. Finally, we demonstrate that changing this new representation space allows us to change the generated style of an already trained generator. World-GAN enables its users to generate Minecraft worlds based on parts of their creations.

show abstract

“…Previous efforts to build artificial worlds include individual machine code programs competing for computational resources [28], [33], simulated creatures in physics simulators [41] [42] [56] [46], games [7], [16], [45], [48], or abstract worlds [44]. I find the open-ended universes based on cellular automata such as SmoothLife, Lenia, and genelife [6], [25], [30] particulary interesting.…”

Section: Related Workmentioning

confidence: 99%

Carle's Game: An Open-Ended Challenge in Exploratory Machine Creativity

Davis

2021

Preprint

View full text Add to dashboard Cite

This paper is both an introduction and an invitation. It is an introduction to CARLE, a Life-like cellular automata simulator and reinforcement learning environment. It is also an invitation to Carle's Game, a challenge in open-ended machine exploration and creativity. Inducing machine agents to excel at creating interesting patterns across multiple cellular automata universes is a substantial challenge, and approaching this challenge is likely to require contributions from the fields of artificial life, AI, machine learning, and complexity, at multiple levels of interest. Carle's Game is based on machine agent interaction with CARLE, a Cellular Automata Reinforcement Learning Environment. CARLE is flexible, capable of simulating any of the 262,144 different rules defining Life-like cellular automaton universes. CARLE is also fast and can simulate automata universes at a rate of tens of thousands of steps per second through a combination of vectorization and GPU acceleration. Finally, CARLE is simple. Compared to high-fidelity physics simulators and video games designed for human players, CARLE's two-dimensional grid world offers a discrete, deterministic, and atomic universal playground, despite its complexity. In combination with CARLE, Carle's Game offers an initial set of agent policies, learning and meta-learning algorithms, and reward wrappers that can be tailored to encourage exploration or specific tasks.

show abstract

EvoCraft: A New Challenge for Open-Endedness

Cited by 15 publications

References 25 publications

Evolution through Large Models

Evolution through Large Models

World-GAN: a Generative Model for Minecraft Worlds

Carle's Game: An Open-Ended Challenge in Exploratory Machine Creativity

Contact Info

Product

Resources

About