Update README.md
Browse files
README.md
CHANGED
|
@@ -3,28 +3,32 @@ model-index:
|
|
| 3 |
- name: twitter-roberta-base-hate-latest
|
| 4 |
results: []
|
| 5 |
pipeline_tag: text-classification
|
|
|
|
|
|
|
| 6 |
---
|
| 7 |
# cardiffnlp/twitter-roberta-base-hate-latest
|
| 8 |
|
| 9 |
This model is a fine-tuned version of [cardiffnlp/twitter-roberta-base-2022-154m](https://huggingface.co/cardiffnlp/twitter-roberta-base-2022-154m) for binary hate-speech classification. A combination of 13 different hate-speech datasets in the English language were used to fine-tune the model.
|
| 10 |
|
| 11 |
## Following metrics are achieved
|
| 12 |
-
|
| 13 |
-
|
|
| 14 |
-
|
|
| 15 |
-
|
|
| 16 |
-
|
|
| 17 |
-
|
|
| 18 |
-
|
|
| 19 |
-
|
|
| 20 |
-
|
|
| 21 |
-
| Twitter
|
| 22 |
-
|
|
| 23 |
-
|
|
| 24 |
-
|
|
| 25 |
-
|
|
| 26 |
-
|
|
| 27 |
-
|
|
|
|
|
|
|
|
| 28 |
|
| 29 |
|
| 30 |
### Usage
|
|
@@ -39,4 +43,17 @@ model = tweetnlp.Classifier("cardiffnlp/twitter-roberta-base-hate-latest")
|
|
| 39 |
model.predict('I love everybody :)')
|
| 40 |
>> {'label': 'NOT-HATE'}
|
| 41 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
```
|
|
|
|
| 3 |
- name: twitter-roberta-base-hate-latest
|
| 4 |
results: []
|
| 5 |
pipeline_tag: text-classification
|
| 6 |
+
language:
|
| 7 |
+
- en
|
| 8 |
---
|
| 9 |
# cardiffnlp/twitter-roberta-base-hate-latest
|
| 10 |
|
| 11 |
This model is a fine-tuned version of [cardiffnlp/twitter-roberta-base-2022-154m](https://huggingface.co/cardiffnlp/twitter-roberta-base-2022-154m) for binary hate-speech classification. A combination of 13 different hate-speech datasets in the English language were used to fine-tune the model.
|
| 12 |
|
| 13 |
## Following metrics are achieved
|
| 14 |
+
|
| 15 |
+
| **Dataset** | **Accuracy** | **Macro-F1** | **Weighted-F1** |
|
| 16 |
+
|:----------|-----------:|-----------:|--------------:|
|
| 17 |
+
| hatEval, SemEval-2019 Task 5: Multilingual Detection of Hate Speech Against Immigrants and Women in Twitter | 0.5831 | 0.5646 | 0.548 |
|
| 18 |
+
| ucberkeley-dlab/measuring-hate-speech | 0.9273 | 0.9193 | 0.928 |
|
| 19 |
+
| Detecting East Asian Prejudice on Social Media | 0.9231 | 0.6623 | 0.9428 |
|
| 20 |
+
| Call me sexist, but | 0.9686 | 0.9203 | 0.9696 |
|
| 21 |
+
| Predicting the Type and Target of Offensive Posts in Social Media | 0.9164 | 0.6847 | 0.9098 |
|
| 22 |
+
| HateXplain | 0.8653 | 0.845 | 0.8662 |
|
| 23 |
+
| Large Scale Crowdsourcing and Characterization of Twitter Abusive BehaviorLarge Scale Crowdsourcing and Characterization of Twitter Abusive Behavior | 0.7801 | 0.7446 | 0.7614 |
|
| 24 |
+
| Multilingual and Multi-Aspect Hate Speech Analysis | 0.9944 | 0.4986 | 0.9972 |
|
| 25 |
+
| Hate speech and offensive content identification in indo-european languages | 0.8779 | 0.6904 | 0.8706 |
|
| 26 |
+
| Are You a Racist or Am I Seeing Things? | 0.921 | 0.8935 | 0.9216 |
|
| 27 |
+
| Automated Hate Speech Detection | 0.9423 | 0.9249 | 0.9429 |
|
| 28 |
+
| Hate Towards the Political Opponent | 0.8783 | 0.6595 | 0.8788 |
|
| 29 |
+
| Hateful Symbols or Hateful People? | 0.8187 | 0.7833 | 0.8323 |
|
| 30 |
+
| **Overall** | **0.8766** | **0.7531** | **0.8745** |
|
| 31 |
+
|
| 32 |
|
| 33 |
|
| 34 |
### Usage
|
|
|
|
| 43 |
model.predict('I love everybody :)')
|
| 44 |
>> {'label': 'NOT-HATE'}
|
| 45 |
|
| 46 |
+
```
|
| 47 |
+
|
| 48 |
+
|
| 49 |
+
### Model based on:
|
| 50 |
+
```
|
| 51 |
+
@misc{antypas2023robust,
|
| 52 |
+
title={Robust Hate Speech Detection in Social Media: A Cross-Dataset Empirical Evaluation},
|
| 53 |
+
author={Dimosthenis Antypas and Jose Camacho-Collados},
|
| 54 |
+
year={2023},
|
| 55 |
+
eprint={2307.01680},
|
| 56 |
+
archivePrefix={arXiv},
|
| 57 |
+
primaryClass={cs.CL}
|
| 58 |
+
}
|
| 59 |
```
|