DynaSteer dynamically steers LLM reasoning trajectories toward truth via pattern clustering, Fisher-LDA projection, and entropy-triggered representation edits, improving performance on MATH and generalizing to coding.
arXiv preprint arXiv:2508.04530 , year=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Search for Truth from Reasoning: A Dynamic Representation Editing Framework for Steering LLM Trajectories
DynaSteer dynamically steers LLM reasoning trajectories toward truth via pattern clustering, Fisher-LDA projection, and entropy-triggered representation edits, improving performance on MATH and generalizing to coding.