ONNX
ONNX: An open format for representing models and transferring them between frameworks and inference runtimes. The graph describes operators and tensors, while different runtimes optimize execution for specific hardware. Operator coverage and numerical differences must be checked on the actual model.
In italiano: ONNXLegge IV · Loss, gradiente e ottimizzatoriThe graph describes operators and tensors, while different runtimes optimize execution for specific hardware. Operator coverage and numerical differences must be checked on the actual model.
Examples
- Export from PyTorch
- ONNX Runtime
- Cross-platform deployment