|
--- |
|
|
|
language: |
|
- en |
|
tags: |
|
- fast |
|
- coreference-resolution |
|
license: mit |
|
datasets: |
|
- multi_news |
|
- ontonotes |
|
metrics: |
|
- CoNLL |
|
task_categories: |
|
- coreference-resolution |
|
model-index: |
|
- name: biu-nlp/f-coref |
|
results: |
|
- task: |
|
type: coreference-resolution |
|
name: coreference-resolution |
|
dataset: |
|
name: ontonotes |
|
type: coreference |
|
metrics: |
|
- name: Avg. F1 |
|
type: CoNLL |
|
value: 78.5 |
|
|
|
--- |
|
|
|
## F-Coref: Fast, Accurate and Easy to Use Coreference Resolution |
|
|
|
[F-Coref](https://arxiv.org/abs/2209.04280) allows to process 2.8K OntoNotes documents in 25 seconds on a V100 GPU (compared to 6 minutes for the [LingMess](https://arxiv.org/abs/2205.12644) model, and to 12 minutes of the popular AllenNLP coreference model) with only a modest drop in accuracy. |
|
The fast speed is achieved through a combination of distillation of a compact model from the LingMess model, and an efficient batching implementation using a technique we call leftover |
|
|
|
Please check the [official repository](https://github.com/shon-otmazgin/fastcoref) for more details and updates. |
|
|
|
#### Experiments |
|
|
|
| Model | Runtime | Memory | |
|
|-----------------------|---------|---------| |
|
| [Joshi et al. (2020)](https://arxiv.org/abs/1907.10529) | 12:06 | 27.4 | |
|
| [Otmazgin et al. (2022)](https://arxiv.org/abs/2205.12644) | 06:43 | 4.6 | |
|
| + Batching | 06:00 | 6.6 | |
|
| [Kirstain et al. (2021)](https://arxiv.org/abs/2101.00434) | 04:37 | 4.4 | |
|
| [Dobrovolskii (2021)](https://arxiv.org/abs/2109.04127) | 03:49 | 3.5 | |
|
| [F-Coref](https://arxiv.org/abs/2209.04280) | 00:45 | 3.3 | |
|
| + Batching | 00:35 | 4.5 | |
|
| + Leftovers batching | 00:25 | 4.0 | |
|
The inference time(Min:Sec) and memory(GiB) for each model on 2.8K documents. Average of 3 runs. Hardware, NVIDIA Tesla V100 SXM2. |
|
|
|
### Citation |
|
|
|
``` |
|
@inproceedings{otmazgin-etal-2022-f, |
|
title = "{F}-coref: Fast, Accurate and Easy to Use Coreference Resolution", |
|
author = "Otmazgin, Shon and |
|
Cattan, Arie and |
|
Goldberg, Yoav", |
|
booktitle = "Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing: System Demonstrations", |
|
month = nov, |
|
year = "2022", |
|
address = "Taipei, Taiwan", |
|
publisher = "Association for Computational Linguistics", |
|
url = "https://aclanthology.org/2022.aacl-demo.6", |
|
pages = "48--56", |
|
abstract = "We introduce fastcoref, a python package for fast, accurate, and easy-to-use English coreference resolution. The package is pip-installable, and allows two modes: an accurate mode based on the LingMess architecture, providing state-of-the-art coreference accuracy, and a substantially faster model, F-coref, which is the focus of this work. F-coref allows to process 2.8K OntoNotes documents in 25 seconds on a V100 GPU (compared to 6 minutes for the LingMess model, and to 12 minutes of the popular AllenNLP coreference model) with only a modest drop in accuracy. The fast speed is achieved through a combination of distillation of a compact model from the LingMess model, and an efficient batching implementation using a technique we call leftover batching. https://github.com/shon-otmazgin/fastcoref", |
|
} |
|
``` |
