Pith. sign in

arXiv preprint arXiv:2509.12886 , year=

The LLM Already Knows: Estimating LLM- Perceived Question Difficulty via Hidden Representations · arXiv 2509.12886

3 Pith papers cite this work. Polarity classification is still indexing.

3 Pith papers citing it
read on arXiv browse 3 citing papers

fields

cs.CL 2 cs.CY 1

years

2026 3

verdicts

UNVERDICTED 2 CONDITIONAL 1

representative citing papers

Sparse Reward Subsystem in Large Language Models

cs.CL · 2026-02-01 · unverdicted · novelty 6.0

LLM hidden states contain a sparse reward subsystem consisting of value neurons that predict state value and dopamine neurons that encode step-level temporal difference errors.

Future Confidence Distillation in Large Language Models

cs.CL · 2026-07-08 · conditional · novelty 5.0

Linear probes trained on pre-solution hidden states, supervised by post-solution correctness probe outputs, recover 32–66% of the calibration gap between pre- and post-solution confidence across five open-source LLMs.

citing papers explorer

Showing 3 of 3 citing papers.

  • Cognitive offloading and the speedup illusion in human-AI interaction cs.CY · 2026-05-22 · unverdicted · none · ref 73

    Preregistered behavioral study identifies a speedup illusion where users overestimate time savings from AI assistance on cognitive tasks despite no actual difference in completion times.

  • Sparse Reward Subsystem in Large Language Models cs.CL · 2026-02-01 · unverdicted · none · ref 26

    LLM hidden states contain a sparse reward subsystem consisting of value neurons that predict state value and dopamine neurons that encode step-level temporal difference errors.

  • Future Confidence Distillation in Large Language Models cs.CL · 2026-07-08 · conditional · none · ref 21

    Linear probes trained on pre-solution hidden states, supervised by post-solution correctness probe outputs, recover 32–66% of the calibration gap between pre- and post-solution confidence across five open-source LLMs.