Following instructions in real-world conditions requires a capability to adapt to the world’s volatility and entanglement: the environment is dynamic and unpredictable, instructions can be linguistically complex with diverse vocabulary, and the number of possible goals an agent may encounter is vast. Despite extensive research in this area, most studies are conducted in static environments with simple instructions and a limited vocabulary, making it difficult to assess agent performance in more diverse and challenging settings. To address this gap, we introduce CrafText, a benchmark for evaluating instruction following in a multimodal environment with diverse instructions and dynamic interactions. CrafText includes 3,924 instructions with 3,423 unique words, covering Localization, Conditional, Building, and Achievement tasks. Additionally, we propose an evaluation protocol that measures an agent’s ability to generalize to novel instruction formulations and dynamically evolving task configurations, providing a rigorous test of both linguistic understanding and adaptive decision-making.
DOI: 10.18653/v1/2025.acl-long.1267
Скачать статью (PDF) из архива конференции ACL (англ.): https://aclanthology.org/2025.acl-long.1267/
Скачать статью (PDF) на arXiv.org (англ.): https://arxiv.org/abs/2505.11962
Скачать код на GitHub: https://github.com/AIRI-Institute/CrafText
Zoya Volovikova, Gregory Gorbov, Petr Kuderov, Aleksandr Panov, and Alexey Skrynnik. 2025. CrafText Benchmark: Advancing Instruction Following in Complex Multimodal Open-Ended World // In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 26131–26151, Vienna, Austria.