A Practical Approach for Few Shot Learning with SetFit - Haystack EU 2023

Explore a practical approach to few-shot learning using SetFit (Sentence Transformer Fine Tuning) in this 28-minute conference talk from Haystack EU 2023. Discover how SetFit achieves state-of-the-art results for classification problems in label-scarce settings, outperforming even GPT-3 in many cases. Learn about a case study in the legal research domain, where a small dataset of real-world search queries with relevant and irrelevant results was used to train a model using SetFit. Understand how word embeddings were generated for semantic searching and how a ranking model was trained and compared with other approaches. Gain insights into SetFit's ranking application and the results of the experiments conducted. Presented by Fernando Vieira da Silva, CEO of N2VEC and Ph.D. in Computer Science with expertise in Natural Language Processing, this talk offers valuable knowledge for those interested in advanced search techniques and relevance ranking in enterprise document search.