Text Encoders Lack Knowledge: Leveraging Generative LLMs for Domain-Specific Semantic Textual Similarity
Abstract
STS can be cast as a text generation problem while maintaining strong performance on multiple STS benchmarks.
Type
Publication
The 2023 Conference on Empirical Methods in Natural Language Processing (Generation, Evaluation & Metrics Workshop)
Parker Seegmiller
Innovation Fellow, PhD Candidate @ Dartmouth CS | Applied Science Intern @ Amazon
nlp @ dartmouth college