metadata
dataset_info:
features:
- name: content
dtype: string
splits:
- name: train
num_bytes: 134025361660
num_examples: 19450435
download_size: 68402599157
dataset_size: 134025361660
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
size_categories:
- 10M<n<100M
This is a filtered subset of the allenai/peS2o dataset which has been:
- Stripped of references using a combination of regexes.
- Aggressively minhash deduped using a shingle size of 20 and a threshold of 0.5.
- Run through a Lingua English language filter with a confidence threshold of 0.99.
- Run through an OSCAR trained English KenLM model (at
edugp/kenlm) perplexity filter for values between 8 and 550.