dataset_info:
features:
- name: epoch
dtype: float64 - name: step
dtype: int64 - name: batch_size
dtype: int64 - name: mean
dtype: float64 - name: std
dtype: float64 - name: min
dtype: float64 - name: p10
dtype: float64 - name: median
dtype: float64 - name: p90
dtype: float64 - name: max
dtype: float64 - name: pos_frac
dtype: float64 - name: sample
sequence: float64 - name: npy
dtype: string
splits: - name: train
num_bytes: 460056
num_examples: 661
download_size: 382274
dataset_size: 460056
configs: - config_name: default
data_files:- split: train
path: data/train-*
- split: train