README.md

# intent-onnx

This is an ONNX version of the fine-tuned intent classification model.

## Model Details

- **Model Type**: BERT-based sentence transformer
- **Architecture**: BertModel
- **Hidden Size**: 384
- **Attention Heads**: 12
- **Layers**: 3
- **Vocabulary Size**: 30,522

## Usage

### Transformers.js

```javascript
import { pipeline } from '@xenova/transformers';

const extractor = await pipeline('feature-extraction', 'drithh/intent-onnx');

const output = await extractor('your text here', {
  pooling: 'mean',
  normalize: true
});

console.log('Embedding shape:', output.data.length);
```

### Python

```python
from sentence_transformers import SentenceTransformer

model = SentenceTransformer('drithh/intent-classifier')
embeddings = model.encode('your text here')
print(f'Embedding shape: {embeddings.shape}')
```

## Training

This model was fine-tuned on intent classification data using SentenceTransformers.