Sample-Efficient Post-Training for LEGO Spatial-Physics Reasoning
LLM-based LEGO assembly generation requires both semantic grounding and physical feasibility. We identify a data-induced failure mode, PhysHack, in which the assemblies satisfy physical-validity constraints while producing structures that are geometrically misaligned, semantically inconsistent, or poorly calibrated. To address this challenge, we propose a model-based data selection approach that uses only a small fraction of the training data while improving physically grounded LEGO assembly gen
Record details
Published: 29 May 2026
Source: arXiv
Category: Research
Topics: Safety & alignment
Retrieved: 14 July 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
Contribution Weights: A Geometrical Analysis of Self-Attention Transformers
arXiv · 29 May 2026
A Persona-Based Evaluation Framework for Pluralistic Alignment in Generative AI
arXiv · 29 May 2026
ImmersiveTTS: Environment-Aware Text-to-Speech with Multimodal Diffusion Transformer and Domain-Specific Representation Alignment
arXiv · 29 May 2026
MIMO: Multilingual Information Retrieval via Monolingual Objectives
arXiv · 29 May 2026
TUX: Measuring Human--AI Tacit Understanding
arXiv · 29 May 2026
What Makes LVLMs Hallucinate Less? Unveiling the Architectural Factors Behind Hallucination Robustness
arXiv · 29 May 2026
How to cite this record
ethics.ai (29 May 2026), “Sample-Efficient Post-Training for LEGO Spatial-Physics Reasoning,” evidence record 3445, https://ethics.ai/record/3445 (originally published by arXiv).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.