README.md

metadata

dataset_info:
  features:
    - name: image
      dtype: image
    - name: camera_pose
      list: float32
      length: 3
    - name: primitive_types
      list: int32
    - name: primitive_thetas
      list:
        list: float32
        length: 6
    - name: file_id
      dtype: string
    - name: corners
      list:
        list: float64
  splits:
    - name: train
      num_bytes: 340625803
      num_examples: 15583
    - name: validation
      num_bytes: 41969124
      num_examples: 1948
    - name: test
      num_bytes: 40891591
      num_examples: 1948
  download_size: 421983371
  dataset_size: 423486518
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*

Quick Start (One-Line Load)

from datasets import load_dataset

# 1. Load directly from the Hugging Face Hub
dataset = load_dataset("heoy567/abc_lines", split="train")

splits = dataset.train_test_split(test_size=0.2, seed=42)
train_data = splits["train"]
test_data = splits["test"]