metadata
dataset_info:
features:
- name: latent
dtype: int64
- name: count
dtype: float64
- name: total
dtype: float64
- name: mean
dtype: float64
- name: layer_mean
dtype: float64
- name: layer_var
dtype: float64
- name: layer_std
dtype: float64
- name: counts
sequence: float64
- name: totals
sequence: float64
- name: means
sequence: float64
- name: probs
sequence: float64
splits:
- name: train
num_bytes: 22413312
num_examples: 49152
download_size: 14521317
dataset_size: 22413312
configs:
- config_name: default
data_files:
- split: train
path: data/train-*