This is an earlier version. View the current version.

Merged Abilities in an Environment with New Transitions

Authors

  • Gregory Lawson Author
  • Austin Harrison Author
  • Joseph Franklin Author

Abstract

Cross-modality merging can retain verbal success while an agent lacks a grounded world model in a new environment. Evaluation should preserve modality-specific outcomes and independently check transitions produced by exploration. A static aggregate answer score cannot reveal every failure in action consequences. Parameter combination makes an intervention available without demonstrating its interactive benefit. This focused synthesis develops an empirical comparison that preserves the terminal action, observation process, and total verification cost. Its proposed tests are prospective; no new simulation run, model-training result, or intervention effect is reported.

References

  1. S. Chen, T. Zhu, Z. Wang, J. Zhang, K. Wang, R. Zhou, S. Gao, T. Xiao, Y. W. Teh, J. He, and M. Li, “Why Do LLM Agents Fail in Exploring New Environments? A World-Modeling Perspective,” Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, 2026.
  2. S. Chen, J. Zhang, T. Zhu, W. Liu, S. Gao, M. Xiong, M. Li, and J. He, “Bring Reason to Vision: Understanding Perception and Reasoning through Model Merging,” Proceedings of the 42nd International Conference on Machine Learning, 2025.
  3. Zhou, Yongxi, Junwei Yao, Yuanzhe Liu, Zihan Dong, Wenbo Ye, Jiaxi Wen, and Lai Yun Choi. “Single Canonical Prompts Underestimate LLM Safety's Surface-Form Sensitivity.” arXiv, 2026, arXiv:2608.02665. https://arxiv.org/abs/2608.02665.
  4. Rasmussen, Carl Edward, and Christopher K. I. Williams. Gaussian Processes for Machine Learning. MIT Press, 2006. https://gaussianprocess.org/gpml/.
  5. Hu, Edward J., et al. “LoRA: Low-Rank Adaptation of Large Language Models.” International Conference on Learning Representations, 2022. https://openreview.net/forum?id=nZeVKeeFYf9.
  6. Guo, Chuan, Geoff Pleiss, Yu Sun, and Kilian Q. Weinberger. “On Calibration of Modern Neural Networks.” Proceedings of the 34th International Conference on Machine Learning, PMLR, vol. 70, 2017, pp. 1321–1330. https://proceedings.mlr.press/v70/guo17a.html.
  7. Sutton, Richard S., and Andrew G. Barto. Reinforcement Learning: An Introduction. 2nd ed., MIT Press, 2018. http://incompleteideas.net/book/the-book-2nd.html.
  8. Levine, Sergey, Aviral Kumar, George Tucker, and Justin Fu. “Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems.” arXiv, 2020, arXiv:2005.01643. https://arxiv.org/abs/2005.01643.
  9. Kumar, Aviral, Aurick Zhou, George Tucker, and Sergey Levine. “Conservative Q-Learning for Offline Reinforcement Learning.” Advances in Neural Information Processing Systems, vol. 33, 2020, pp. 1179–1191. https://arxiv.org/abs/2006.04779.
  10. Lewis, Patrick, et al. “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks.” Advances in Neural Information Processing Systems, vol. 33, 2020, pp. 9459–9474. https://arxiv.org/abs/2005.11401.
  11. Geifman, Yonatan, and Ran El-Yaniv. “Selective Classification for Deep Neural Networks.” Advances in Neural Information Processing Systems, vol. 30, 2017. https://arxiv.org/abs/1705.08500.

Downloads

Published

2026-10-12

Versions

Issue

Vol. 2 No. 10 (2026): October 2026

Section

Articles