|
|
---
|
|
|
license: apache-2.0
|
|
|
datasets:
|
|
|
- dair-ai/emotion
|
|
|
language:
|
|
|
- en
|
|
|
metrics:
|
|
|
- accuracy
|
|
|
- f1
|
|
|
- precision
|
|
|
- recall
|
|
|
base_model:
|
|
|
- albert/albert-large-v2
|
|
|
pipeline_tag: text-classification
|
|
|
---
|
|
|
# Sentiment classification using Albert-large-v2 |
|
|
|
|
|
### Model Description |
|
|
|
|
|
This model is a fine-tuned version of the ALBERT-Large model designed for **emotion sentiment classification**. This model is capable of detecting six different emotional categories in text: **Anger**, **Disgust**, **Fear**, **Happiness**, **Sadness**, and **Surprise**. It achieves high performance on sentiment classification tasks, making it suitable for a variety of real-world applications such as emotion detection, content moderation, and sentiment analysis. |
|
|
|
|
|
## How to Get Started |
|
|
|
|
|
Use the code below to get started with the model. |
|
|
|
|
|
```python |
|
|
from transformers import pipeline |
|
|
|
|
|
emotion_classifier = pipeline("text-classification", model="SandeepVvigneshwar/sentiment-classification-albert-large-v2") |
|
|
|
|
|
text = "I am so happy to be part of this project!" |
|
|
emotion = emotion_classifier(text) |
|
|
print(emotion) |
|
|
``` |
|
|
|
|
|
|
|
|
## Requirements |
|
|
|
|
|
- Python 3.x |
|
|
- Hugging Face `transformers` library |
|
|
- PyTorch or TensorFlow |
|
|
|
|
|
|
|
|
### Training Data |
|
|
|
|
|
[dair-ai/emotion](https://huggingface.co/datasets/dair-ai/emotion) |
|
|
|
|
|
#### Training Hyperparameters |
|
|
|
|
|
- learning_rate = 2e-5 |
|
|
- per_device_train_batch_size = 8 |
|
|
- per_device_eval_batch_size = 8 |
|
|
- gradient_accumulation_steps = 2 |
|
|
- num_train_epochs = 8 |
|
|
- weight_decay = 0.01 |
|
|
- fp16 = True |
|
|
- metric_for_best_model = "f1" |
|
|
- dataloader_num_workers = 4 |
|
|
- max_grad_norm = 1.0 |
|
|
- lr_scheduler_type = "linear" |
|
|
|
|
|
### Limits |
|
|
|
|
|
- Domain-specific Text: The model may not perform well on specialized or highly technical texts. |
|
|
- Languages: The model has been fine-tuned on English-language data and may not generalize well to other languages. |
|
|
- Input Length: The model performs best with shorter text inputs. For longer, more complex texts, performance may vary. |
|
|
|
|
|
## Evaluation |
|
|
|
|
|
| Metric | Value | |
|
|
|----------------------------|--------| |
|
|
| **Evaluation Loss** | 0.08795 | |
|
|
| **Evaluation Accuracy** | 94.31% | |
|
|
| **Evaluation F1-Score** | 94.39% | |
|
|
| **Evaluation Precision** | 94.99% | |
|
|
| **Evaluation Recall** | 94.31% | |
