igorsterner
commited on
Commit
•
cd3bc7e
1
Parent(s):
f852680
Upload README.md
Browse files
README.md
ADDED
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- multilingual
|
4 |
+
- en
|
5 |
+
- de
|
6 |
+
license: mit
|
7 |
+
widget:
|
8 |
+
- text: "ich glaub ich muss echt rewatchen like i [MASK] so empty was soll ich denn jetzt machen"
|
9 |
+
example_title: "Example 1"
|
10 |
+
- text: "I don't get [MASK] er damit erreichen will."
|
11 |
+
example_title: "Example 2"
|
12 |
+
- text: "Sagt ein(e) Head(in) [MASK] research! Researchen Sie mal ein bisschen mehr."
|
13 |
+
example_title: "Example 3"
|
14 |
+
---
|
15 |
+
|
16 |
+
# German-English Code-Switching BERT
|
17 |
+
|
18 |
+
A BERT-based model trained with masked language modelling on a large corpus of German--English code-switching. It was introduced in [this paper](). This model is case sensitive.
|
19 |
+
|
20 |
+
## Overview
|
21 |
+
- **Initializd language model:** bert-base-multilingual-cased
|
22 |
+
- **Training data:** The TongueSwitcher Corpus
|
23 |
+
- **Infrastructure**: 4x Nvidia A100 GPUs
|
24 |
+
- **Published**: 16 October 2023
|
25 |
+
|
26 |
+
## Hyperparameters
|
27 |
+
|
28 |
+
```
|
29 |
+
batch_size = 32
|
30 |
+
n_steps = 191,950
|
31 |
+
max_seq_len = 512
|
32 |
+
learning_rate = 1e-4
|
33 |
+
weight_decay = 0.01
|
34 |
+
Adam beta = (0.9, 0.999)
|
35 |
+
lr_schedule = LinearWarmup
|
36 |
+
num_warmup_steps = 10,000
|
37 |
+
seed = 2021
|
38 |
+
```
|
39 |
+
|
40 |
+
## Performance
|
41 |
+
|
42 |
+
During training we monitored the evaluation loss on the TongueSwitcher dev set.
|
43 |
+
|
44 |
+
![dev loss](loss.png)
|
45 |
+
|
46 |
+
## Authors
|
47 |
+
- Igor Sterner: `is473 [at] cam.ac.uk`
|
48 |
+
- Simone Teufel: `sht25 [at] cam.ac.uk`
|
49 |
+
|
50 |
+
### BibTeX entry and citation info
|
51 |
+
|
52 |
+
```bibtex
|
53 |
+
@inproceedings{sterner2023tongueswitcher,
|
54 |
+
author = {Igor Sterner and Simone Teufel},
|
55 |
+
title = {TongueSwitcher: Fine-Grained Identification of German-English Code-Switching},
|
56 |
+
booktitle = {Sixth Workshop on Computational Approaches to Linguistic Code-Switching},
|
57 |
+
year = {2023},
|
58 |
+
}
|
59 |
+
```
|