EVOKE: Eliciting World Knowledge in Agents for Transferable Decision-Making
cs.CL
Submitted: 2026-09-29
Updated: 2026-10-01
Comments: 19 pages. Project page: https://gnonymous.github.io/EVOKE ; Code: https://github.com/Gnonymous/EVOKE ; Models: https://huggingface.co/Gnonymous/EVOKE
License: http://creativecommons.org/licenses/by/4.0/
The gist: Large language models (LLMs) are increasingly deployed as agents for multi-step decision-making, yet transfer poorly to unseen environments.
Terminology
Abstract
Large language models (LLMs) are increasingly deployed as agents for multi-step decision-making, yet transfer poorly to unseen environments. World-model methods address this by training agents to predict future observations, at the cost of additional training and errors that compound when predictions are used for planning. However, for LLM agents operating in digital environments, much of this world knowledge is already internalized during pretraining, which shifts the problem from acquiring it to eliciting it. We argue that typical post-training provides little pressure for such elicitation, since supervision under a single goal at each visited state inadvertently drives policies to rely on superficial contextual habits. We introduce EVOKE, a post-training method that supplies this pressure through goal diversity at fixed states. Motivated by theory showing that an agent competent across diverse goals must encode a world model recoverable from its action preferences, EVOKE holds the environment state and interaction history fixed and ranks the same candidate actions under alternative goals, forcing action preferences to change, so that a policy relying on contextual habits or single-goal correlations cannot order them correctly. This implicitly elicits the policy's pretrained world knowledge to inform decisions. We evaluate EVOKE across diverse tasks in three backbones, demonstrating improved task performance, unseen environment generalization, and data efficiency. We further conduct controlled analyses to better understand what drives these gains. These findings offer a new perspective on eliciting internalized world knowledge for transferable action through direct decision supervision.
Sources
- V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
- LoRA: Low-Rank Adaptation of Large Language Models
- AHEAD: Adaptive Hindsight with Environment-Augmented Distillation for Agentic RL
- Imagine-then-Plan: Agent Learning from Adaptive Lookahead with World Models
- Policy and World Modeling Co-Training for Language Agents
- Self-Distilled Agentic Reinforcement Learning
- PCSD: Persistent Consistency for Self-Distillation in Agentic Reinforcement Learning
- Qwen2.5 Technical Report
- Contrastive Learning with Hard Negative Samples
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- ALFWorld: Aligning Text and Embodied Environments for Interactive Learning
- Skill-SD: Skill-Conditioned Self-Distillation for Multi-turn LLM Agents
- Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution
- MemWM: Memory-Augmented Text-Based World Model
- EnvRL: Learn from Environment Dynamics in Agentic Reinforcement Learning
- Qwen3 Technical Report
- Self-Distilled RLVR
- OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning
- Reinforcement World Model Learning for LLM-based Agents
- Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering