metadata
dataset_info:
features:
- name: instruction
dtype: string
- name: input
dtype: string
- name: audio_id
dtype: string
- name: dataset
dtype: string
- name: task
dtype: string
- name: output
dtype: string
- name: audio
struct:
- name: array
sequence: float64
- name: path
dtype: string
- name: sampling_rate
dtype: int64
- name: question
dtype: string
- name: answer
dtype: string
- name: thinking
dtype: string
splits:
- name: train
num_bytes: 8682610023
num_examples: 6881
download_size: 6503871601
dataset_size: 8682610023
configs:
- config_name: default
data_files:
- split: train
path: data/train-*