SemEval Idiom Interpretation Project

SemEval_Admire_UCSC.ipynb

This notebook contains code for processing and analyzing idioms for the SemEval task:

• Loads and preprocesses datasets (test or extended)
• Classifies sentence_type as literal or idiomatic
• Implements two prompting strategies to generate definitions for idioms
• Ranks images for examples using CLIP, OpenCLIP, ALIGN
• Creates submission files for evaluation

prompting_outputs

The prompting_outputs directory contains the outputs for sentence type classification and idiom definition.

Results are organized by dataset (test vs. extended).