Commit
•
f01568f
1
Parent(s):
e8306e6
Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: cc-by-sa-4.0
|
3 |
+
pipeline_tag: fill-mask
|
4 |
+
arxiv: 2210.05529
|
5 |
+
language: en
|
6 |
+
thumbnail: https://github.com/coastalcph/hierarchical-transformers/raw/main/data/figures/hat_encoder.png
|
7 |
+
tags:
|
8 |
+
- long-documents
|
9 |
+
datasets:
|
10 |
+
- c4
|
11 |
+
model-index:
|
12 |
+
- name: kiddothe2b/hierarchical-transformer-base-4096-v2
|
13 |
+
results: []
|
14 |
+
---
|
15 |
+
|
16 |
+
# Hierarchical Attention Transformer (HAT) / hierarchical-transformer-base-4096-v2
|
17 |
+
|
18 |
+
## Disclaimer 🚧 ⚠️
|
19 |
+
This is an experimental version of HAT, trying to make HAT a native part of Transformers library. Please use ONLY [kiddothe2b/hierarchical-transformer-base-4096](https://huggingface.co/kiddothe2b/hierarchical-transformer-base-4096) for the moment.
|
20 |
+
|
21 |
+
## Model description
|
22 |
+
|
23 |
+
This is a Hierarchical Attention Transformer (HAT) model as presented in [An Exploration of Hierarchical Attention Transformers for Efficient Long Document Classification (Chalkidis et al., 2022)](https://arxiv.org/abs/2210.05529).
|
24 |
+
|
25 |
+
The model has been warm-started re-using the weights of RoBERTa (Liu et al., 2019), and continued pre-trained for MLM in long sequences following the paradigm of Longformer released by Beltagy et al. (2020). It supports sequences of length up to 4,096.
|
26 |
+
|
27 |
+
HAT uses hierarchical attention, which is a combination of segment-wise and cross-segment attention operations. You can think of segments as paragraphs or sentences.
|
28 |
+
|
29 |
+
|
30 |
+
## Citing
|
31 |
+
|
32 |
+
If you use HAT in your research, please cite:
|
33 |
+
|
34 |
+
[An Exploration of Hierarchical Attention Transformers for Efficient Long Document Classification](https://arxiv.org/abs/2210.05529). Ilias Chalkidis, Xiang Dai, Manos Fergadiotis, Prodromos Malakasiotis, and Desmond Elliott. 2022. arXiv:2210.05529 (Preprint).
|
35 |
+
|
36 |
+
```
|
37 |
+
@misc{chalkidis-etal-2022-hat,
|
38 |
+
url = {https://arxiv.org/abs/2210.05529},
|
39 |
+
author = {Chalkidis, Ilias and Dai, Xiang and Fergadiotis, Manos and Malakasiotis, Prodromos and Elliott, Desmond},
|
40 |
+
title = {An Exploration of Hierarchical Attention Transformers for Efficient Long Document Classification},
|
41 |
+
publisher = {arXiv},
|
42 |
+
year = {2022},
|
43 |
+
}
|
44 |
+
```
|
45 |
+
|
46 |
+
|