Emotional framing in quantitative reasoning problems reduces LLM accuracy by 2-10 percentage points, recoverable by neutralization, unlike neutral paraphrases.
Solving general arithmetic word problems
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
verdicts
UNVERDICTED 2representative citing papers
MarCos modifies transformers to perform continuous multi-step reasoning by mapping thought-level continuous states directly to next-thought distributions, achieving substantial wall-clock speedups on math problems.
citing papers explorer
-
TEMPER: Testing Emotional Perturbation in Quantitative Reasoning
Emotional framing in quantitative reasoning problems reduces LLM accuracy by 2-10 percentage points, recoverable by neutralization, unlike neutral paraphrases.
-
Deep Thinking by Markov Chain of Continuous Thoughts
MarCos modifies transformers to perform continuous multi-step reasoning by mapping thought-level continuous states directly to next-thought distributions, achieving substantial wall-clock speedups on math problems.