In the coming months, you will also have access to a `model.export()` API for inference-only model export, relying on the SavedModel format. It will include options for target-specific optimizations — think mobile, browser, microcontrollers, etc.

→ View original post on X — @fchollet,