Model Upload Guide

EdgeAI supports a wide array of neural architectures. Learn how to prepare your weights for the sharding engine.

Supported Frameworks

PyTorch (.pt, .pth)
ONNX (.onnx)
TensorFlow (pb)
Keras (.h5)

Ingestion Protocol

01

Neural Integrity Check

The SDK performs a local SHA-256 hash to ensure no data corruption during transmission.

02

Format Normalization

Models are automatically converted to a high-performance internal representation for sharding.

03

Architecture Mapping

We map layer operations to optimized hardware kernels for your specific edge target.

CLI Upload Example

The fastest way to upload is via the terminal. Large models are automatically sharded into 50MB chunks for resilient streaming.

edge-ai upload ./massive_resnet.onnx