PrepBench is a benchmark showing that state-of-the-art LLMs still struggle with natural-language-driven data preparation involving disambiguation, code generation, and workflow translation.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.DB 1years
2026 1verdicts
UNVERDICTED 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
PrepBench: How Far Are We from Natural-Language-Driven Data Preparation?
PrepBench is a benchmark showing that state-of-the-art LLMs still struggle with natural-language-driven data preparation involving disambiguation, code generation, and workflow translation.