Planners Exploit Deterministic Error: Dithered Low-Bit World Models
cs.RO, cs.LG
Submitted: 2026-11-09
Updated: 2026-11-09
Comments: 7 pages, 3 figures, 1 table
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Stochastic Rounding in Low-Precision Transformer Inference: A Variable-Precision Emulation Study of a Small GPT-2
- An Empirical Study of World Model Quantization
- Low-Discrepancy Dither for Quantized Recurrent State Caches
- TD-MPC-Opt: Distilling Model-Based Multi-Task Reinforcement Learning Agents
- When Low Prediction Error Misleads Planning: Diagnosing Representation, Dynamics, and Decision Failures in Latent World Models
- Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
- Pretraining Large Language Models with NVFP4
- World-Model Gradient Refinement in CEM: An Empirical Study
- Recovering Low-Bit Failures in Vision-Language-Action Policies via Quantization-Conditioned Reinforcement Learning
- Where Bits Matter in World Model Planning: A Paired Mixed-Bit Study for Efficient Spatial Reasoning
- Microscaling Data Formats for Deep Learning
- DeepMind Control Suite
- Not All Errors Matter: Decision-Relevant Prediction Error Predicts Planning Quality
- SteerQuant: Steering Quantization Error with Action-Guided Scaling in World-Action Models
- Which Decisions Low-Bit Quantization Breaks, and How to Predict Them
- Predict Before You Deploy: Offline Prediction of Quantization-Induced Task Degradation for World Action Models
- VLAQuantBench: Closed-Loop Evaluation of Post-Training Quantization for Vision-Language-Action Models
- A Control Theory of Predictability in Latent World Models
- DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planning
Related papers
- Rethinking Actions: Continuity-Aware Priors for World-Model Planning
- MOSAIC: Modular Obstacle and Spatial-Aware Framework for Manipulation In Cluttered Scenes
- SAGAS: Semantic-Aware Graph-Assisted Stitching for Offline Temporal Logic Planning
- DexSeed: Expanding Human Hand-Object Interactions into Robot Demonstrations for Dexterous Policy Learning
- LaST0.5: A Generalist World Action Model with Interaction-Centric Latent Reasoning
- CIALLO: Learning Shared Intent Representations for Cross-Embodiment Manipulation