deqing/addition_dataset_4digits
收藏Hugging Face2026-05-27 更新2026-05-31 收录
下载链接:
https://hf-mirror.com/datasets/deqing/addition_dataset_4digits
下载链接
链接失效反馈官方服务:
资源简介:
---
configs:
- config_name: 1BT
data_files:
- split: train
path: 1BT/train-*.parquet
- split: test
path: 1BT/test-*.parquet
---
# 4-digit addition dataset (a + b = c, all <= 4 digits)
Exhaustive enumeration of every (a, b) with a, b in [0, 9999] and
a + b <= 9999 (so c is also <= 4 digits). 50,005,000 ordered pairs,
commutative-safe 90/10 train/test split.
Train ~45M examples (~570M tokens under Llama-3 tokenizer); test ~5M examples.
提供机构:
deqing


