Fine-tuning, multi-agent, and DPO approaches all improve cognitive and linguistic authenticity of LLM-simulated math students over few-shot prompts, with each offering distinct benefits for teacher noticing of student thinking as revealed in interviews.
EdArXiv Preprints (2024),https://osf.io/preprints/edarxiv/5zwv3_v1
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.HC 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
LLMs achieve highest dialogue annotation accuracy via multi-agent prompting but show context-dependent performance and directional biases, with better results in K-12 affective coding and systematic errors in cognitive and behavioral categories.
citing papers explorer
-
Developing Authentic Simulated Learners for Mathematics Teacher Learning: Insights from Three Approaches with Large Language Models
Fine-tuning, multi-agent, and DPO approaches all improve cognitive and linguistic authenticity of LLM-simulated math students over few-shot prompts, with each offering distinct benefits for teacher noticing of student thinking as revealed in interviews.
-
Decoding Student Dialogue: A Multi-Dimensional Comparison and Bias Analysis of Large Language Models as Annotation Tools
LLMs achieve highest dialogue annotation accuracy via multi-agent prompting but show context-dependent performance and directional biases, with better results in K-12 affective coding and systematic errors in cognitive and behavioral categories.