Project information

This application is a search-based retrieval demo that uses GloVe (Global Vectors for Word Representation) and Sentence Transformers to find the closest words to a given input. The user can select a model type and input categories and a sentence. The application then checks if the GloVe embeddings exist for the selected model type. If not, it downloads them. The application then loads the GloVe embeddings and calculates the cosine similarity for the input sentence using both GloVe and Sentence Transformers. The results are displayed to the user, showing the closest words to the input sentence according to the categories provided. The results are also visualized in a chart. The GloVe model used in this application is pretrained on 2 billion tweets with a vocabulary size of 1.2 million.