JeanMax Voice Assistant Model

JeanMax is a PyTorch-based neural network model for voice task intent classification and conversational response generation. It's designed for voice assistant applications that need to understand user commands and generate appropriate responses.

Model Architecture

  • Input: Text vectorized using TF-IDF with n-gram features (unigrams, bigrams, trigrams)
  • Architecture: Multi-task deep neural network with:
    • 2 fully connected layers (256 → 128 neurons)
    • Layer normalization and ReLU activation
    • Dropout for regularization
    • Dual output heads: Intent classification + Response generation
  • Output:
    • Intent classification (command understanding)
    • Response generation (conversational replies)

Usage

from huggingface_hub import hf_hub_download
import torch

# Download model
model_path = hf_hub_download("sachaniyanarvin/Jeanmax_vs", "JeanMax.pt")
config_path = hf_hub_download("sachaniyanarvin/Jeanmax_vs", "config.json")

# Load model
checkpoint = torch.load(model_path, map_location='cpu')
model = JeanMaxNeuralNet(
    input_dim=checkpoint['input_dim'],
    num_intents=checkpoint['num_intents'],
    num_responses=checkpoint['num_responses']
)
model.load_state_dict(checkpoint['model_state_dict'])
model.eval()

# Make prediction
text = "open chrome"
vector = vectorizer.transform(text)
input_tensor = torch.tensor(vector, dtype=torch.float32).unsqueeze(0)

with torch.no_grad():
    intent_logits, response_logits = model(input_tensor)
    intent_pred = torch.argmax(intent_logits, dim=1)

Training

  • Framework: PyTorch
  • Training Data: Command corpus + conversational corpus
  • Loss Function: Multi-task loss (CrossEntropy for intent + response)
  • Optimizer: AdamW with weight decay
  • Training Mode: Supports transfer learning for new intents

Intents Supported

The model supports various voice commands including:

  • Application control (open, close applications)
  • System operations (system status, tracking)
  • Conversational interactions (greetings, questions)
  • And more...

Model Files

  • JeanMax.pt - PyTorch model checkpoint with weights and metadata
  • config.json - Model configuration
  • model.py - Model architecture definition
  • requirements.txt - Python dependencies

License

MIT License

Author

Created by Narvin Sachaniya for the Friday Voice Assistant project.

Downloads last month
8
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support