Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn natural language processing techniques for building semantic search systems and working with vector databases through this comprehensive 9-hour course. Master dense vector representations for NLP and computer vision applications, then dive deep into sentence embeddings using transformer models. Explore multiple fine-tuning approaches including the original NLI Softmax loss method and the high-performance Multiple Negatives Ranking technique. Discover how to work with multilingual sentence vectors that support over 50 languages using a single model, and understand unsupervised sentence transformer training through TSDAE (Transformer-based Sequential Denoising Auto-Encoder). Examine evaluation measures for search and recommender systems, then advance to data augmentation techniques with Augmented SBERT and domain transfer methods using AugSBERT. Build practical question-answering systems by implementing both extractive and abstractive QA approaches, including complete Python implementations for open-domain question-answering and reader models. Learn advanced training techniques such as query generation (GenQ) for sentence transformers and explore cutting-edge methods like Generative Pseudo-Labeling (GPL) that may represent the future of sentence transformer development.
