Explore and manage STM32 ML models with the STM32AI Model Zoo dashboard
Quantize a model for faster inference
Evaluate reward models for math reasoning
Measure execution times of BERT models using WebGPU and WASM
Evaluate code generation with diverse feedback types
View and submit machine learning model evaluations
Upload a machine learning model to Hugging Face Hub
Track, rank and evaluate open LLMs and chatbots
Explore and benchmark visual document retrieval models
Evaluate and submit AI model results for Frugal AI Challenge
Measure BERT model performance using WASM and WebGPU
Compare LLM performance across benchmarks
Display model benchmark results
The STM32 Model Zoo app is a comprehensive dashboard designed for exploring, managing, and benchmarking machine learning models optimized for STM32 microcontrollers. It provides a centralized platform for users to discover pre-trained models, evaluate performance metrics, and deploy them efficiently. This tool is essential for developers working with STM32 devices to streamline their ML model development and deployment process.
• Model Browser: Easily explore and filter models based on criteria like model size, complexity, and performance.
• Performance Metrics: View detailed benchmark results, including inference time, memory usage, and accuracy.
• Model Comparison: Compare multiple models side-by-side to identify the best fit for specific applications.
• Export Functionality: Directly export models in formats compatible with STM32 devices.
• Integration Support: Compatible with STM32Cube.AI tools for seamless integration into existing workflows.
• User-Friendly Interface: Intuitive design with visualizations to simplify model selection and management.
What are the system requirements for the STM32 Model Zoo app?
The app is designed to work with STM32Cube.AI tools and requires a compatible operating system (e.g., Windows, Linux). Ensure you have the latest STM32Cube.AI software installed.
How do I find models suited for my specific application?
Use the model browser’s filters to narrow down models by size, complexity, or application domain. You can also search by keyword or sort by performance metrics.
Can I integrate custom models into the STM32 Model Zoo app?
Yes, you can add custom models to the Model Zoo by exporting them in the required format (e.g., .zip file containing model weights and metadata) and uploading them through the app’s interface.