---
language: en
license: mit
tags:
- shopping
- deals
- classification
- e-commerce
- recommendation
datasets:
- custom
metrics:
- accuracy
- precision
- recall
- f1
widget:
- text: I'm looking for headphones
  example_title: Electronics Search
- text: Do you have any kitchen appliance deals?
  example_title: Kitchen Deals
- text: Show me the best laptop deals
  example_title: Laptop Deals
- text: I need a new smart TV
  example_title: Smart TV Search
library_name: transformers
---

# Shopping Assistant Model

This model is trained to classify shopping deals into categories. It can be used to provide better recommendations to users based on their queries.

## Model description

The model is based on DistilBERT and is fine-tuned on a dataset of shopping deals. It is trained to classify deals into the following categories:
electronics, clothing, home, kitchen, toys, other

## Intended uses & limitations

This model is intended to be used for classifying shopping deals and providing recommendations to users. It is not intended to be used for any other purpose.

## Training data

The model is trained on a dataset of shopping deals from DealsFinders.com. The dataset is automatically labeled based on keywords in the deal titles and descriptions.

## Training procedure

The model is trained using the Hugging Face Transformers library. It is fine-tuned on the deals dataset using a multi-label classification approach.

## Evaluation results

The model is evaluated on a held-out validation set. The evaluation metrics include accuracy, precision, recall, and F1 score.

## Usage

```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

# Load the model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("selvaonline/shopping-assistant")
model = AutoModelForSequenceClassification.from_pretrained("selvaonline/shopping-assistant")

# Prepare the input
text = "I'm looking for headphones"
inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)

# Get the model prediction
with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.sigmoid(outputs.logits)

# Load the categories
categories = ["electronics", "clothing", "home", "kitchen", "toys", "other"]

# Get the top categories
top_categories = []
for i, score in enumerate(predictions[0]):
    if score > 0.5:  # Threshold for multi-label classification
        top_categories.append((categories[i], score.item()))

# Sort by score
top_categories.sort(key=lambda x: x[1], reverse=True)

# Print the results
for category, score in top_categories:
    print(f"{category}: {score:.4f}")
```