Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,28 @@
|
|
1 |
---
|
2 |
license: cc-by-4.0
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: cc-by-4.0
|
3 |
---
|
4 |
+
This is model based on mT5-L that predicts a binary label for a given article and summary for Q5 (main idea(s)), as defined in the [SEAHORSE paper](https://arxiv.org/abs/2305.13194) (Clark et al., 2023).
|
5 |
+
|
6 |
+
It is trained similarly to the [TRUE paper (Honovich et al, 2022)](https://arxiv.org/pdf/2204.04991.pdf) on human ratings from the SEAHORSE dataset in 6 languages:
|
7 |
+
- German
|
8 |
+
- English
|
9 |
+
- Spanish
|
10 |
+
- Russian
|
11 |
+
- Turkish
|
12 |
+
- Vietnamese
|
13 |
+
|
14 |
+
The input format for the model is: "premise: ARTICLE hypothesis: SUMMARY".
|
15 |
+
|
16 |
+
There is also an XXL version of this model, as well as metrics trained for each of the other 5 dimensions described in the original paper.
|
17 |
+
|
18 |
+
The full citation for the SEAHORSE paper is:
|
19 |
+
```
|
20 |
+
@misc{clark2023seahorse,
|
21 |
+
title={SEAHORSE: A Multilingual, Multifaceted Dataset for Summarization Evaluation},
|
22 |
+
author={Elizabeth Clark and Shruti Rijhwani and Sebastian Gehrmann and Joshua Maynez and Roee Aharoni and Vitaly Nikolaev and Thibault Sellam and Aditya Siddhant and Dipanjan Das and Ankur P. Parikh},
|
23 |
+
year={2023},
|
24 |
+
eprint={2305.13194},
|
25 |
+
archivePrefix={arXiv},
|
26 |
+
primaryClass={cs.CL}
|
27 |
+
}
|
28 |
+
```
|