From 2213f7df9e33656f3e2070d2fed907afaa547eaa Mon Sep 17 00:00:00 2001 From: Aditya Date: Wed, 23 Oct 2024 17:33:52 -0700 Subject: [PATCH] Added Note for YOLO models --- manual/working_with_models.rst | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/manual/working_with_models.rst b/manual/working_with_models.rst index 0ecefba..02db605 100644 --- a/manual/working_with_models.rst +++ b/manual/working_with_models.rst @@ -27,6 +27,12 @@ The following network model formats are supported: PyTorch models can be saved in different formats, only `TorchScript` format is supported. see :ref:`working-with-pytorch-models-label` for more information. +.. note:: + + For YOLO models in ``Tensorflow Lite`` format, we recommend selecting the ``float 32`` (fp32) model format. + SyNAP will quantize and optimize this format to run efficiently on the NPU. + + For YOLO models in ``ONNX`` format, we recommend using the ``kl_divergence`` algorithm to optimize prediction results. .. note::