Abstract: We introduce ADASTT, an adaptive meta-learning framework that selects, in real time, the most suitable speech-to-text (STT) model for each incoming audio input. Factors such as background ...
Abstract: In Retrieval-Augmented Generation (RAG), large language models (LLMs) typically utilize the top-k contexts returned by a retriever or employ instruction fine-tuning frameworks to fine-tune ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results