LLM fine-tuned on Qwen 3.5 estimates item parameters using simulated response probabilities

Researchers have developed a method to estimate item parameters for multiple-choice models using a fine-tuned large language model based on Qwen 3.5. This approach leverages the LLM's ability to replicate choice probabilities from text and image stimuli, effectively capturing underlying response probabilities by learning from systematic student error patterns. The model demonstrated accuracy in approximating item difficulty on a held-out test set by directly predicting option probabilities. AI

IMPACT This research could improve educational assessment tools by enabling more accurate item parameter estimation through LLM capabilities.

RANK_REASON The cluster contains an academic paper published on arXiv detailing a new methodology. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Christopher Ormerod
• Hugging Face
• Multimodal Item Parameter Estimation using Simulated Response Probabilities
• Qwen 3.5

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →