|
|
--- |
|
|
library_name: transformers |
|
|
tags: |
|
|
- nlp |
|
|
- text |
|
|
- multiclass |
|
|
- classification |
|
|
license: mit |
|
|
base_model: |
|
|
- distilbert/distilbert-base-uncased |
|
|
--- |
|
|
|
|
|
# Model Card for `debojit01/course-review-sentiment` |
|
|
|
|
|
## Model Details |
|
|
|
|
|
**Model Name**: Course Review Sentiment Classifier |
|
|
**Model Type**: Text Classification (Multiclass – Positive, Neutral, Negative) |
|
|
**Language**: English |
|
|
**License**: MIT |
|
|
**Finetuned From**: `distilbert-base-uncased` |
|
|
**Developed By**: Debojit Choudhury |
|
|
|
|
|
## Model Description |
|
|
|
|
|
This model is a fine-tuned DistilBERT model for sentiment classification of course reviews. It predicts whether a review is **positive**, **neutral**, or **negative**, and was trained on a labeled dataset of 100k Coursera reviews. |
|
|
|
|
|
## Uses |
|
|
|
|
|
### Direct Use |
|
|
This model can be used to: |
|
|
- Automatically classify course reviews based on sentiment. |
|
|
- Analyze customer feedback for online education platforms. |
|
|
|
|
|
### Out-of-Scope Use |
|
|
- Not suitable for non-English text. |
|
|
- Not suitable for other domains beyond course review sentiment. |
|
|
|
|
|
## How to Get Started with the Model |
|
|
|
|
|
```python |
|
|
from transformers import pipeline |
|
|
|
|
|
classifier = pipeline("text-classification", model="debojit01/course-review-sentiment") |
|
|
classifier("The course was extremely helpful and well-structured!") |
|
|
``` |
|
|
|
|
|
## Training Details |
|
|
|
|
|
### Training Data |
|
|
|
|
|
Kaggle's 100k Coursera Reviews Dataset |
|
|
|
|
|
- **Number of Classes**: 3 |
|
|
- **Training Framework**: Hugging Face Transformers |
|
|
- **Max Seq Length**: 512 |
|
|
- **Epochs**: 3 |
|
|
|
|
|
## Evaluation |
|
|
|
|
|
- **Test Split**: 20% of full dataset |
|
|
- **Metrics**: Accuracy, Macro Precision, Recall, F1 |
|
|
- **Macro F1**: 0.7647813475266324 |
|
|
- **Accuracy**: 0.7641242937853108 |
|
|
- **Macro Precision**: 0.766738569737377 |
|
|
- **Macro Recall**: 0.7641242937853107 |
|
|
- **samples_per_second**: 72.966 |
|
|
- **steps_per_second**: 1.159 |
|
|
|
|
|
## Environmental Impact |
|
|
|
|
|
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly --> |
|
|
|
|
|
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). |
|
|
|
|
|
- **Hardware Type:** Tesla T4 GPU (Google Colab) |
|
|
- **Hours used:** <2 hours |
|
|
- **Compute Region:** US (Colab) |
|
|
|
|
|
## Citation |
|
|
|
|
|
If you use this model, please cite: |
|
|
|
|
|
Debojit Choudhury, Course Review Sentiment Classifier (2025), Hugging Face. |
|
|
https://huggingface.co/debojit01/course-review-sentiment |
