lilonghao/Audio-Cogito

Audio-Cogito: Towards Deep Audio Reasoning in Large Audio Language Models Audio-Cogito is a large-scale audio reasoning dataset introduced in the paper Audio-Cogito: Towards Deep Audio Reasoning in Large Audio Language Models. The released data contains 545k high-quality audio reasoning samples spanning sound, speech, and music domains. Each sample includes label annotations, Chain-of-Thought (CoT) annotations, and final answers. Links… See the full description on the dataset page: https://huggingface.co/datasets/lilonghao/Audio-Cogito.