README.md

---
tags:
- text-classification
- emotion-analysis
language:
- it
widget:
- text: I love AutoTrain 🤗
datasets:
- tradicio/autotrain-data-it-emotion-analysis
- dair-ai/emotion
co2_eq_emissions:
  emissions: 0.4489187526120041
license: cc-by-sa-4.0
metrics:
- accuracy
- f1
- recall
pipeline_tag: text-classification
---
# IT-EMOTION-ANALYZER

This is a model for emotion analysis of italian sentences trained on a translated dataset by [Google Translator](https://pypi.org/project/deep-translator/). It maps sentences & paragraphs with 6 emotions which are:

- 0: sadness
- 1: joy
- 2: love
- 3: anger
- 4: fear
- 5: surprise

<!--- Describe your model here -->

## Model in action

Using this model becomes easy when you have [transformers](https://github.com/huggingface/transformers) installed:

```
pip install -U transformers
```

Then you can use the model like this:

```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
from transformers import pipeline

sentences = ["Questa è una frase triste", "Questa è una frase felice", "Questa è una frase di stupore"]

tokenizer = AutoTokenizer.from_pretrained("aiknowyou/it-emotion-analyzer")
model = AutoModelForSequenceClassification.from_pretrained("aiknowyou/it-emotion-analyzer")

emotion_analysis = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)
emotion_analysis(sentences)
```
Obtaining the following result:
```python
[{'label': '0', 'score': 0.9481984972953796},
 {'label': '1', 'score': 0.9299975037574768},
 {'label': '5', 'score': 0.9543816447257996}]
```

# Model Trained Using AutoTrain

- Problem type: Multi-class Classification
- Model ID: 43095109829
- CO2 Emissions (in grams): 0.4489

## Validation Metrics

- Loss: 0.566
- Accuracy: 0.828
- Macro F1: 0.828
- Micro F1: 0.828
- Weighted F1: 0.828
- Macro Precision: 0.828
- Micro Precision: 0.828
- Weighted Precision: 0.828
- Macro Recall: 0.828
- Micro Recall: 0.828
- Weighted Recall: 0.828