emnist_mnist / README.md
tanganke's picture
Convert dataset to Parquet (#1)
193c42b verified
---
dataset_info:
config_name: emnist-mnist
features:
- name: image
dtype: image
- name: label
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
splits:
- name: train
num_bytes: 25332252.0
num_examples: 60000
- name: test
num_bytes: 4230499.0
num_examples: 10000
download_size: 28641083
dataset_size: 29562751.0
configs:
- config_name: emnist-mnist
data_files:
- split: train
path: emnist-mnist/train-*
- split: test
path: emnist-mnist/test-*
default: true
---
# Dataset Card for "emnist-mnist"
## Dataset Information
The `emnist-mnist` dataset is a set of images of handwritten digits. The dataset is split into a training set and a test set.
## Data Fields
- `image`: The image of the handwritten digit. The data type of this field is `image`.
- `label`: The label of the handwritten digit. The data type of this field is `class_label`, and it can take on the values '0' to '9'.
## Data Splits
- `train`: The training set consists of 60000 examples, with a total size of 25579752 bytes.
- `test`: The test set consists of 10000 examples, with a total size of 4271749 bytes.