# InnerThoughts: Disentangling Representations and Predictions in Large Language Models

Source: [InnerThoughts: Disentangling Representations and Predictions in Large Language Models](<https://www.sota2.com/research/paper/68fa76ce1e209131cea6f2c4>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T04:02:16.872Z

Authors: Didier Ch\\'etelat, Joseph Cotnareanu, Rylee Thompson, Yingxue Zhang, Mark Coates

Published: 2025-01-29T21:01:44.000Z

## Abstract

Large language models (LLMs) contain substantial factual knowledge which is commonly elicited by multiple-choice question-answering prompts. Internally, such models process the prompt through multiple transformer layers, building varying representations of the problem within its hidden states. Ultimately, however, only the hidden state corresponding to the final layer and token position are used to predict the answer label. In this work, we propose instead to learn a small separate neural network predictor module on a collection of training questions, that take the hidden states from all the layers at the last temporal position as input and outputs predictions. In effect, such a framework disentangles the representational abilities of LLMs from their predictive abilities. On a collection of hard benchmarks, our method achieves considerable improvements in performance, sometimes comparable to supervised fine-tuning procedures, but at a fraction of the computational cost.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2501.17994>)

[PDF](<https://arxiv.org/pdf/2501.17994.pdf>)
