Datasets:

text-machine-lab
/

vocab_filtered_dataset_2.1B

Modalities:

Formats:

Size:

ArXiv:

Libraries:

Dataset card Data Studio Files Files and versions Community

SherinB commited on Apr 2, 2024

Commit

47600d9

·

verified ·

1 Parent(s): 7dca0f1

Update README.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -24,7 +24,7 @@ configs:
 ### Dataset Summary
 This data is the simplified vocabulary-filtered pretraining data published by "Emergent Abilities in Reduced-Scale Generative Language Models". The vocabulary is derived from the AO-Childes speech corpus (https://github.com/UIUCLearningLanguageLab/AOCHILDES)
-We filter the train split of SlimPajama dataset (https://huggingface.co/datasets/cerebras/SlimPajama-627B) based on the AO-Childes vocabulary retaining spans which contain integers, symbols, and words that belong to the AO-Childes vocabulary. Around 1.5% of Out of Vocabulary words are also allowed.
 ### Citation Information
 If this dataset is useful to you please cite our work.

 ### Dataset Summary
 This data is the simplified vocabulary-filtered pretraining data published by "Emergent Abilities in Reduced-Scale Generative Language Models". The vocabulary is derived from the AO-Childes speech corpus (https://github.com/UIUCLearningLanguageLab/AOCHILDES)
+We filter the train split of SlimPajama dataset (https://huggingface.co/datasets/cerebras/SlimPajama-627B) based on the AO-Childes vocabulary retaining spans which contain integers, symbols, and words that belong to the AO-Childes vocabulary. Around 1.5% of Out of Vocabulary words are also allowed. A contiguous span of 100 tokens are selected.
 ### Citation Information
 If this dataset is useful to you please cite our work.