Get Started
Home
Topics
Search
Library
Research questionHow can language models infer statistical tasks and relevant variable roles from informal goals and heterogeneous data?Users often describe what they want informally while providing heterogeneous data, so the intended statistical task and useful variables may not be explicit. A language model must resolve both decisions rather than assume the analysis target is already specified.
AI
Evaluation & Benchmarks
Machine Learning
Natural Language Processing
Statistical Machine Learning
Latest papersRecent research connected to this question, newest first.Benchmarking Language Models for Statistical Problem FormulationThe evidence concerns zero-shot evaluation of 14 open- and closed-source language models on StatFormBench, built from five cross-domain statistics textbooks and a data science case library. The benchmark contains 1,013 samples covering 20 coarse-grained and 85 fine-grained statistical problem categories, and measures task classification and variable-set overlap; reported best scores are 72.0% fine-grained classification accuracy and 63.2% variable-set overlap, with no model consistently best across both subtasks.research paper · Sep 5, 2026
Related questions
Can language models infer others’ mental states as social interactions evolve under unreliable information?What language should a language model produce when its training data are translated, synthetic, or heavily curated?How can noisy LLM conditionals support coherent probabilistic inference over structured variables without order-dependent generation?How should low-resource LLM fine-tuning use task-level language priors with ambiguous or incomplete data?