convertToEngineFromUff with int8 mode when Meeting "uniformly zero, network calibration failed"

fwx5 · November 17, 2018, 5:24am

When I did convertToEngineFromUff with int8 mode, I met following error. How can I fix it? My model can run successfully in fp32 mode.

ERROR: Tensor FeatureExtractor/InceptionV2/Mixed_4a/Branch_1/Conv2d_0a_1x1/Relu6/relu2 is uniformly zero; network calibration failed.

AastaLLL · November 19, 2018, 1:34am

Hi,

Please generate a calibration table for your model.

When creating networks in 8-bit integer precision, it uses a process called calibration to determine the dynamic range of intermediate activations, and hence the appropriate scaling factors for quantization.

Here are more information for your reference:

Introduction for INT8: [url]http://on-demand.gputechconf.com/gtc/2017/presentation/s7310-8-bit-inference-with-tensorrt.pdf[/url]
Calibration tutorial: [url]https://devblogs.nvidia.com/int8-inference-autonomous-vehicles-tensorrt/[/url]

Thanks.

Topic		Replies	Views
tensorrt int8 calibration failed for a tensor is uniformly zero. TensorRT	6	2052	October 12, 2021
ERROR: Tensor is uniformly zero; network calibration failed. TensorRT	3	1430	January 25, 2019
Int8 Calibration failing when one layer's output is uniformly zero TensorRT	4	1484	October 12, 2021
Inferencing on AGX Xavier in INT8 mode Jetson AGX Xavier jetson-inference	3	1060	December 8, 2021
I convert the Onnx model to Int8 model failed！ TensorRT	1	923	January 9, 2020
int8 calibration problem TensorRT	1	1037	January 16, 2019
Converting Caffe to TensorRT using int8 Jetson TX2	4	777	October 18, 2021
Int8 inference driver error Jetson AGX Xavier jetson-inference	5	361	October 18, 2021
ERROR:Calibration failure occured with no scaling factors detected TensorRT	0	2415	March 12, 2019
TensorRT Python INT8 calibration failure TensorRT	3	1935	November 23, 2018

convertToEngineFromUff with int8 mode when Meeting "uniformly zero, network calibration failed"

Related topics