Commit
•
838aa00
1
Parent(s):
6057067
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: es
|
| 3 |
+
license: mit
|
| 4 |
+
|
| 5 |
+
widget:
|
| 6 |
+
- text: "y porqué es lo que hay que hacer con los menas y con los adultos también!!!! NO a los inmigrantes ilegales!!!!"
|
| 7 |
+
---
|
| 8 |
+
|
| 9 |
+
### Description
|
| 10 |
+
This model is a fine-tuned version of [BETO (spanish bert)](https://huggingface.co/dccuchile/bert-base-spanish-wwm-uncased) that has been trained on the *Datathon Against Racism* dataset (2022)
|
| 11 |
+
|
| 12 |
+
We performed several experiments that will be described in the upcoming paper "Estimating Ground Truth in a Low-labelled Data Regime:A Study of Racism Detection in Spanish" (NEATClasS 2022)
|
| 13 |
+
We applied 6 different methods ground-truth estimations, and for each one we performed 4 epochs of fine-tuning. The result is made of 24 models:
|
| 14 |
+
|
| 15 |
+
| method | epoch 1 | epoch 3 | epoch 3 | epoch 4 |
|
| 16 |
+
|--- |--- |--- |--- |--- |
|
| 17 |
+
| raw-label | [raw-label-epoch-1](https://huggingface.co/MartinoMensio/racism-models-raw-label-epoch-1) | [raw-label-epoch-2](https://huggingface.co/MartinoMensio/racism-models-raw-label-epoch-2) | [raw-label-epoch-3](https://huggingface.co/MartinoMensio/racism-models-raw-label-epoch-3) | [raw-label-epoch-4](https://huggingface.co/MartinoMensio/racism-models-raw-label-epoch-4) |
|
| 18 |
+
| m-vote-strict | [m-vote-strict-epoch-1](https://huggingface.co/MartinoMensio/racism-models-m-vote-strict-epoch-1) | [m-vote-strict-epoch-2](https://huggingface.co/MartinoMensio/racism-models-m-vote-strict-epoch-2) | [m-vote-strict-epoch-3](https://huggingface.co/MartinoMensio/racism-models-m-vote-strict-epoch-3) | [m-vote-strict-epoch-4](https://huggingface.co/MartinoMensio/racism-models-m-vote-strict-epoch-4) |
|
| 19 |
+
| m-vote-nonstrict | [m-vote-nonstrict-epoch-1](https://huggingface.co/MartinoMensio/racism-models-m-vote-nonstrict-epoch-1) | [m-vote-nonstrict-epoch-2](https://huggingface.co/MartinoMensio/racism-models-m-vote-nonstrict-epoch-2) | [m-vote-nonstrict-epoch-3](https://huggingface.co/MartinoMensio/racism-models-m-vote-nonstrict-epoch-3) | [m-vote-nonstrict-epoch-4](https://huggingface.co/MartinoMensio/racism-models-m-vote-nonstrict-epoch-4) |
|
| 20 |
+
| regression-w-m-vote | [regression-w-m-vote-epoch-1](https://huggingface.co/MartinoMensio/racism-models-regression-w-m-vote-epoch-1) | [regression-w-m-vote-epoch-2](https://huggingface.co/MartinoMensio/racism-models-regression-w-m-vote-epoch-2) | [regression-w-m-vote-epoch-3](https://huggingface.co/MartinoMensio/racism-models-regression-w-m-vote-epoch-3) | [regression-w-m-vote-epoch-4](https://huggingface.co/MartinoMensio/racism-models-regression-w-m-vote-epoch-4) |
|
| 21 |
+
| w-m-vote-strict | [w-m-vote-strict-epoch-1](https://huggingface.co/MartinoMensio/racism-models-w-m-vote-strict-epoch-1) | [w-m-vote-strict-epoch-2](https://huggingface.co/MartinoMensio/racism-models-w-m-vote-strict-epoch-2) | [w-m-vote-strict-epoch-3](https://huggingface.co/MartinoMensio/racism-models-w-m-vote-strict-epoch-3) | [w-m-vote-strict-epoch-4](https://huggingface.co/MartinoMensio/racism-models-w-m-vote-strict-epoch-4) |
|
| 22 |
+
| w-m-vote-nonstrict | [w-m-vote-nonstrict-epoch-1](https://huggingface.co/MartinoMensio/racism-models-w-m-vote-nonstrict-epoch-1) | [w-m-vote-nonstrict-epoch-2](https://huggingface.co/MartinoMensio/racism-models-w-m-vote-nonstrict-epoch-2) | [w-m-vote-nonstrict-epoch-3](https://huggingface.co/MartinoMensio/racism-models-w-m-vote-nonstrict-epoch-3) | [w-m-vote-nonstrict-epoch-4](https://huggingface.co/MartinoMensio/racism-models-w-m-vote-nonstrict-epoch-4) |
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
This model is `m-vote-strict-epoch-4`
|
| 26 |
+
|
| 27 |
+
### Usage
|
| 28 |
+
|
| 29 |
+
```python
|
| 30 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
|
| 31 |
+
|
| 32 |
+
model_name = 'm-vote-strict-epoch-4'
|
| 33 |
+
tokenizer = AutoTokenizer.from_pretrained("dccuchile/bert-base-spanish-wwm-uncased")
|
| 34 |
+
full_model_path = f'MartinoMensio/racism-models-{model_name}'
|
| 35 |
+
model = AutoModelForSequenceClassification.from_pretrained(full_model_path)
|
| 36 |
+
|
| 37 |
+
pipe = pipeline("text-classification", model = model, tokenizer = tokenizer)
|
| 38 |
+
|
| 39 |
+
texts = [
|
| 40 |
+
'y porqué es lo que hay que hacer con los menas y con los adultos también!!!! NO a los inmigrantes ilegales!!!!',
|
| 41 |
+
'Es que los judíos controlan el mundo'
|
| 42 |
+
]
|
| 43 |
+
print(pipe(texts))
|
| 44 |
+
# [{'label': 'racist', 'score': 0.9965864419937134}, {'label': 'racist', 'score': 0.6058831214904785}]
|
| 45 |
+
```
|
| 46 |
+
|
| 47 |
+
For more details, see https://github.com/preyero/neatclass22
|