---
language:
- en
---

# Text Classification GoEmotions

This model is a fined-tuned version of [nreimers/MiniLMv2-L6-H384-distilled-from-BERT-Large](https://huggingface.co./nreimers/MiniLMv2-L6-H384-distilled-from-BERT-Large) on the on the [Jigsaw 1st Kaggle competition](https://www.kaggle.com/competitions/jigsaw-toxic-comment-classification-challenge) dataset using [unitary/toxic-bert](https://huggingface.co./unitary/toxic-bert) as teacher model.
The quantized version in ONNX format can be found [here](minuva/MiniLMv2-toxic-jijgsaw-onnx). The model with two labels only (toxicity and severe toxicity) is [here](minuva/MiniLMv2-toxic-jijgsaw-lite-onnx)

# Load the Model

```py
from transformers import pipeline

pipe = pipeline(model='minuva/MiniLMv2-toxic-jijgsaw', task='text-classification')
pipe("This is pure trash")
# [{'label': 'toxic', 'score': 0.9383478164672852}]
```

# Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 6e-05
- train_batch_size: 48
- eval_batch_size: 48
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- warmup_ratio: 0.1


# Metrics (comparison with teacher model)

| Teacher (params)    |   Student (params)     | Set  (metric)     | Score (teacher)    |    Score (student)      |
|--------------------|-------------|----------|--------| --------|
| unitary/toxic-bert (110M) |  MiniLMv2-L6-H384-goemotions-v2 (23M)  | Test (ROC_AUC)  | 0.98636 |  0.98600 |

# Deployment

Check this [repository](https://github.com/minuva/toxicity-prediction-serverless) to see how to easily deploy this model in a serverless environment with fast CPU inference and light resource utilization.