metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: completion
dtype: string
- name: states
sequence: int64
- name: problem_states
sequence: int64
- name: contexts
sequence: int64
- name: actions
sequence: int64
- name: text
dtype: string
splits:
- name: train
num_bytes: 114570920
num_examples: 200000
- name: train_rl
num_bytes: 5732666
num_examples: 10000
- name: test
num_bytes: 2349104
num_examples: 4096
download_size: 30404858
dataset_size: 122652690
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: train_rl
path: data/train_rl-*
- split: test
path: data/test-*