Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
Yanli Zhou, Brenden M
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
Attribute retrieval in LLMs follows non-contiguous, redundant layer paths identified via iterative patching, implying highly distributed knowledge storage.
LLMs solve compositional factual recall either by computing intermediates or directly, with mechanism choice correlated to translation geometry in embedding spaces.
Across 15 LLMs, first-answer accuracy on arithmetic procedural execution falls from 63% at 5 steps to 20% at 95 steps, with under-execution increasing.
A survey organizing techniques to achieve efficient reasoning in LLMs by shortening chain-of-thought outputs.
citing papers explorer
-
Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
-
Factual Retrieval in LLMs Is a Redundant, Distributed and Non-Contiguous Process
Attribute retrieval in LLMs follows non-contiguous, redundant layer paths identified via iterative patching, implying highly distributed knowledge storage.
-
How Do Language Models Compose Functions?
LLMs solve compositional factual recall either by computing intermediates or directly, with mechanism choice correlated to translation geometry in embedding spaces.
-
When LLMs Stop Following Steps: A Diagnostic Study of Procedural Execution in Language Models
Across 15 LLMs, first-answer accuracy on arithmetic procedural execution falls from 63% at 5 steps to 20% at 95 steps, with under-execution increasing.
-
Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models
A survey organizing techniques to achieve efficient reasoning in LLMs by shortening chain-of-thought outputs.