The Mouth is Not the Brain: Bridging Energy-Based World Models and Language Generation
cs.LG, cs.AI, cs.CL
Submitted: 2026-01-23
Updated: 2026-03-31
Comments: ICLR 2026 The 2nd Workshop on World Models: Understanding, Modelling, and Scaling
Journal ref: Trans.Mach.Learn.Res.(2026)
License: http://creativecommons.org/licenses/by/4.0/
The gist: Large Language Models (LLMs) generate fluent text, yet whether they truly understand the world or merely produce plausible texts about it remains contested.
Terminology
Abstract
Large Language Models (LLMs) generate fluent text, yet whether they truly understand the world or merely produce plausible texts about it remains contested. We propose an architectural principle, the mouth is not the brain, that explicitly separates world models from language models. Our architecture comprises three components: a DBM that captures domain structure as an energy-based world model, an adapter that projects latent belief states into embedding space, and a frozen GPT-2 that provides linguistic competence without domain knowledge. We instantiate this framework in the consumer review domain using Amazon smartphone reviews. Experiments demonstrate that (1) world model conditioning achieves lower cross-entropy loss and higher semantic similarity than architectural baselines including direct projection and full fine-tuning, while qualitative analysis reveals that soft prompt conditioning resolves a trade-off that prompt-based approaches cannot: simple prompts lack expressiveness while detailed prompts cause output collapse in small LLMs; (2) the DBM's energy function distinguishes coherent from incoherent market configurations, assigning higher energy to implausible brand-price combinations; and (3) interventions on specific attributes propagate causally to generated text with intervened outputs exhibiting distributions statistically consistent with naturally occurring samples sharing the target configuration. These findings suggest that even small-scale language models can achieve consistent, controllable generation when connected to an appropriate world model, providing empirical support for separating linguistic competence from world understanding.
Sources
- World Models
- Hierarchical Memorization in Large Language Models: Evidence from Citation Generation
- Review-LLM: Harnessing Large Language Models for Personalized Review Generation
- On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models
- Hallucination is Inevitable: An Innate Limitation of Large Language Models
- OpenCity: A Scalable Platform to Simulate Urban Activities with Massive LLM Agents
- Fine-Tuning Language Models from Human Preferences
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis