LeNet-5 inference FAIL

[ChaiBapchya]

PR #15783

Problem

LeNet-5 test
Running inference in MXNet
Running inference in MXNet-TensorRT
MXNet accuracy: 99.140000
MXNet-TensorRT accuracy: 99.120000
Run LeNet-5 inference comparison between MXNet and TensorRT. ... FAIL

Build pipeline

http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/mxnet-validation%2Funix-gpu/detail/PR-15783/6/pipeline/

Error Log

======================================================================
FAIL: Run LeNet-5 inference comparison between MXNet and TensorRT.
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/nose/case.py", line 198, in runTest
    self.test(*self.arg)
  File "/work/mxnet/tests/python/tensorrt/test_tensorrt_lenet5.py", line 102, in test_tensorrt_inference
    MXNet = %f, TensorRT = %f""" % (absolute_accuracy_diff, epsilon, mx_pct, trt_pct)
AssertionError: Absolute diff. between MXNet & TensorRT accuracy (0.020000) exceeds threshold (0.010100):
           MXNet = 99.140000, TensorRT = 99.120000
-------------------- >> begin captured logging << --------------------
root: INFO: train-labels-idx1-ubyte.gz exists, skipping download
root: INFO: train-images-idx3-ubyte.gz exists, skipping download
root: INFO: t10k-labels-idx1-ubyte.gz exists, skipping download
root: INFO: t10k-images-idx3-ubyte.gz exists, skipping download
--------------------- >> end captured logging << ---------------------

[mxnet-label-bot]

Hey, this is the MXNet Label Bot.
Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it.
Here are my recommended labels: Bug

[ChaiBapchya]

@mxnet-label-bot add [Flaky]