README.md

---
tags:
- generated_from_keras_callback
model-index:
- name: twitter-roberta-base-sentiment-earthquake
  results: []
---

# twitter-roberta-base-sentiment-earthquake

This is an "extension" of the `twitter-roberta-base-sentiment-latest` [model](https://huggingface.co/cardiffnlp/twitter-roberta-base-sentiment-latest), further finetuned with original Twitter data posted in English about the 10th anniversary of the 2010 Haiti Earthquake.

- Reference Paper: [Sentiment analysis (SA) (supervised and unsupervised classification) of original Twitter data posted in English about the 10th anniversary of the 2010 Haiti Earthquake](https://data.ncl.ac.uk/articles/dataset/Sentiment_analysis_SA_supervised_and_unsupervised_classification_of_original_Twitter_data_posted_in_English_about_the_10th_anniversary_of_the_2010_Haiti_Earthquake/19688040/1). 


## Full classification example

```python
from transformers import AutoModelForSequenceClassification
from transformers import TFAutoModelForSequenceClassification
from transformers import AutoTokenizer
import numpy as np

class_mapping = {0: "Negative", 1: "Neutral", 2: "Positive"}

MODEL = "antypasd/twitter-roberta-base-sentiment-earthquake"

tokenizer = AutoTokenizer.from_pretrained(MODEL)

# PT
model = AutoModelForSequenceClassification.from_pretrained(MODEL)
model.save_pretrained(MODEL)

text = "$202 million of $1.14 billion in United States (US) ​recovery aid went to a new 'industrial park' in Caracol, an area unaffected by the Haiti earthquake. The plan was to invite foreign garment companies to take advantage of extremely low-wage labor"
encoded_input = tokenizer(text, return_tensors='pt')
output = model(**encoded_input)
scores = output[0][0].detach().numpy()
prediction = np.argmax(scores)


# # TF
# model = TFAutoModelForSequenceClassification.from_pretrained(MODEL)
# model.save_pretrained(MODEL)

# encoded_input = tokenizer(text, return_tensors='tf')
# output = model(encoded_input)
# scores = output[0][0].numpy()
# prediction = np.argmax(scores)

# Print label
print(class_mapping[prediction])

```

Output: 

```
Negative
```