| --- |
| language: en |
| license: apache-2.0 |
| tags: |
| - text-classification |
| - sentiment-analysis |
| - distilbert |
| - imdb |
| - huggingface |
| datasets: |
| - imdb |
| metrics: |
| - accuracy |
| --- |
| |
| # DistilBERT Sentiment Analysis (IMDB) |
|
|
| This model is a fine-tuned version of **DistilBERT-base-uncased** for **binary sentiment classification** on movie reviews. |
|
|
| It classifies reviews as **Positive** or **Negative**. |
|
|
| ## Model Details |
|
|
| - **Base Model**: distilbert-base-uncased |
| - **Dataset**: IMDB Movie Reviews (25k train, 25k test) |
| - **Task**: Sentiment Analysis (Positive / Negative) |
| - **Training**: Fine-tuned using Hugging Face Trainer API |
|
|
| ## Results |
|
|
| - **Accuracy on small subset (2k examples)**: **89.0%** |
| - Expected accuracy on full dataset: **~92-94%** |
|
|
| ## Usage |
|
|
| ```python |
| from transformers import pipeline |
| |
| classifier = pipeline( |
| "sentiment-analysis", |
| model="kckdeepak/imdb-distilbert-sentiment-analysis" |
| ) |
| |
| result = classifier("This movie was fantastic!") |
| print(result) |
