Masked Autoencoders that Listen
Audio-MAE sets new SotA on six audio and speech classification tasks, outperforming other recent models that use external supervised pre-training.
Audio-MAE can be used for self-supervised representation learning from audio spectrograms, providing better performance on audio and speech classification tasks. This approach can be implemented to optimize audio and speech analysis in business operations.
Re2G: Retrieve, Rerank, Generate
Large gains in four diverse tasks: zero-shot slot filling, question answering, fact checking and dialog, with relative gains of 9% to 34% over the previous SotA on the KILT leaderboard.
Re2G can be used for retrieval, reranking and generation of text, enhancing performance in tasks such as slot filling, question answering, fact checking, and dialog. This can be implemented in various businesses to optimize natural language processing.