Datasets:

hoskinson-center
/

proof-pile

Tasks:

Text Generation

Modalities:

Sub-tasks:

language-modeling

Languages:

Size:

Tags:

formal-mathematics

Libraries:

License:

Dataset card Viewer Files Files and versions Community

zhangirazerbayev commited on Feb 16, 2023

Commit

664e7c5

•

1 Parent(s): 72455b2

Update README.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -22,7 +22,7 @@ task_ids:
 ---
 # Dataset Description
-The `proof-pile` is a 36GB pre-training dataset of mathematical text that comprises 8.3 billion tokens (using the `gpt2` tokenizer). Models trained on this dataset are coming soon :) The dataset is composed of diverse sources of both informal and formal mathematics, namely
 - ArXiv.math (35GB)
 - Open-source math textbooks (50MB)
 - Formal mathematics libraries (500MB)

 ---
 # Dataset Description
+The `proof-pile` is a 13GB pre-training dataset of mathematical text that comprises 8.3 billion tokens (using the `gpt-neox` tokenizer). Models trained on this dataset are coming soon :) The dataset is composed of diverse sources of both informal and formal mathematics, namely
 - ArXiv.math (35GB)
 - Open-source math textbooks (50MB)
 - Formal mathematics libraries (500MB)