metadata
language:
- en
license: cc-by-sa-4.0
dataset_info:
- config_name: max_len-1024
features:
- name: text
dtype: string
splits:
- name: train
num_bytes: 270688438
num_examples: 449722
download_size: 148512885
dataset_size: 270688438
- config_name: max_len-448
features:
- name: text
dtype: string
splits:
- name: train
num_bytes: 272285366
num_examples: 930858
download_size: 151141635
dataset_size: 272285366
configs:
- config_name: max_len-1024
data_files:
- split: train
path: max_len-1024/train-*
- config_name: max_len-448
data_files:
- split: train
path: max_len-448/train-*
Wikipedia simple splitted
Wikipedia simple data splitted using Langchain's RecursiveCharacterTextSplitter
Usage
- This dataset is meant to be an ultra high quality dataset.
- This can be used for Annealing LLMs.
Why its different
- This dataset is split with a max length of 448 (1283.5) and 1024 (2564) characters
- And rather than splitting by length, it is split using RecursiveCharacterTextSplitter. So, the chunks don't end randomly.
- Can use very large batch sizes.