metadata
dataset_info:
features:
- name: input
struct:
- name: asr
dtype: string
- name: input_json
struct:
- name: messages
list:
- name: message
dtype: string
- name: user
dtype: string
- name: llm
dtype: string
- name: ocr
dtype: string
- name: output_type
dtype: string
- name: possible_values
sequence: string
- name: question
dtype: string
- name: label
dtype: string
splits:
- name: train
num_bytes: 93360
num_examples: 96
- name: validation
num_bytes: 93360
num_examples: 96
download_size: 42674
dataset_size: 186720
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*