metadata
dataset_info:
features:
- name: prompt_hidden_state
sequence: float32
- name: rew_diff
sequence: float64
splits:
- name: train
num_bytes: 35999744
num_examples: 2176
- name: test
num_bytes: 4003648
num_examples: 242
download_size: 15007086
dataset_size: 40003392
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*