|
|
--- |
|
|
language: en |
|
|
license: apache-2.0 |
|
|
tags: |
|
|
- sentiment-analysis |
|
|
- text-classification |
|
|
- transformers |
|
|
- distilbert |
|
|
datasets: |
|
|
- imdb |
|
|
metrics: |
|
|
- accuracy |
|
|
model-index: |
|
|
- name: DistilBERT IMDb Sentiment Classifier |
|
|
results: |
|
|
- task: |
|
|
name: Sentiment Analysis |
|
|
type: text-classification |
|
|
dataset: |
|
|
name: IMDb |
|
|
type: imdb |
|
|
metrics: |
|
|
- name: Accuracy |
|
|
type: accuracy |
|
|
value: 0.88 |
|
|
--- |
|
|
# 🧠 Sentiment Analysis Model — DistilBERT Fine-Tuned on IMDb 🎬 |
|
|
This model is a fine-tuned version of [`distilbert-base-uncased`](https://huggingface.co/distilbert-base-uncased) on the [IMDb movie review dataset](https://huggingface.co/datasets/imdb) for **binary sentiment classification** (positive/negative). It was trained using Hugging Face Transformers and PyTorch. |
|
|
|
|
|
## 🔍 Intended Use |
|
|
|
|
|
This model is designed to classify movie reviews (or other English text) as **positive** or **negative** sentiment. It's ideal for: |
|
|
- Opinion mining |
|
|
- Social media analysis |
|
|
- Review classification |
|
|
- Text classification demos |
|
|
|
|
|
## 🧪 Example Usage |
|
|
|
|
|
```python |
|
|
from transformers import AutoTokenizer, AutoModelForSequenceClassification |
|
|
import torch |
|
|
|
|
|
model_name = "bmdavis/my-language-model" |
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name) |
|
|
model = AutoModelForSequenceClassification.from_pretrained(model_name) |
|
|
|
|
|
text = "This movie was amazing and really well-acted!" |
|
|
inputs = tokenizer(text, return_tensors="pt") |
|
|
outputs = model(**inputs) |
|
|
prediction = torch.argmax(outputs.logits).item() |
|
|
|
|
|
print("Sentiment:", "Positive" if prediction == 1 else "Negative") |
|
|
|
|
|
📊 Dataset |
|
|
IMDb Dataset |
|
|
|
|
|
25,000 training samples |
|
|
|
|
|
25,000 test samples |
|
|
|
|
|
Labels: 0 = Negative, 1 = Positive |
|
|
|
|
|
🧠 Model Details |
|
|
Base Model: distilbert-base-uncased |
|
|
|
|
|
Architecture: Transformer (BERT-like) |
|
|
|
|
|
Framework: PyTorch |
|
|
|
|
|
Tokenizer: WordPiece |
|
|
|
|
|
🛠️ Training |
|
|
Epochs: 3 |
|
|
|
|
|
Batch Size: 8 |
|
|
|
|
|
Optimizer: AdamW |
|
|
|
|
|
Loss: CrossEntropy |
|
|
|
|
|
Trainer API used |
|
|
|
|
|
🔐 License |
|
|
This model is released under the Apache 2.0 license. |
|
|
|
|
|
✍️ Author |
|
|
Created by Brody Davis (@bmdavis) |
|
|
Trained and uploaded using Hugging Face Hub and Transformers |
|
|
|
