metadata
dataset_info:
features:
- name: task_name
dtype: string
- name: task_id
dtype: string
- name: input_ids
list: int64
- name: loss_mask
list: int64
- name: length
dtype: int64
- name: assistant_tokens
dtype: int64
splits:
- name: train
num_bytes: 505365469
num_examples: 2133
- name: validation
num_bytes: 57250077
num_examples: 237
download_size: 47970336
dataset_size: 562615546
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*