Commit
•
2ebcd8f
1
Parent(s):
46b3e16
Add README
Browse files
README.md
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# ScholarBERT_100 Model
|
| 2 |
+
|
| 3 |
+
This is the **ScholarBERT_100** variant of the ScholarBERT model family.
|
| 4 |
+
|
| 5 |
+
The model is pertrained on a large collection of scientific research articles (221B tokens).
|
| 6 |
+
|
| 7 |
+
The model is based on the same architecture as [BERT-large](https://huggingface.co/bert-large-cased) and has a total of 340M parameters.
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
# Model Architecture
|
| 11 |
+
|
| 12 |
+
| Hyperparameter | Value |
|
| 13 |
+
|-----------------|:-------:|
|
| 14 |
+
| Layers | 24 |
|
| 15 |
+
| Hidden Size | 1024 |
|
| 16 |
+
| Attention Heads | 16 |
|
| 17 |
+
| Total Parameters | 340M |
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
# Training Dataset
|
| 21 |
+
|
| 22 |
+
The vocab and the model are pertrained on **100% of the PRD** scientific literature dataset.
|
| 23 |
+
|
| 24 |
+
The PRD dataset is provided by Public.Resource.Org, Inc. (“Public Resource”),
|
| 25 |
+
a nonprofit organization based in California. This dataset was constructed from a corpus
|
| 26 |
+
of journal article files, from which We successfully extracted text from 75,496,055 articles from 178,928 journals.
|
| 27 |
+
The articles span across Arts & Humanities, Life Sciences & Biomedicine, Physical Sciences,
|
| 28 |
+
Social Sciences, and Technology. The distribution of articles is shown below.
|
| 29 |
+
|