diff options
| author | Gunes Bayir <gunes.bayir@arm.com> | 2021-08-04 12:45:33 +0100 |
|---|---|---|
| committer | Gunes Bayir <gunes.bayir@arm.com> | 2021-08-04 14:45:38 +0000 |
| commit | 38ce29d5dce9a09769b55b2717d105f984247e31 (patch) | |
| tree | 2efbed1caa028a18467fc492151f0c44d2559114 /src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_indirect.cpp | |
| parent | 663419457b02238687cb329afcddc73719bdb8fa (diff) | |
| download | ComputeLibrary-38ce29d5dce9a09769b55b2717d105f984247e31.tar.gz | |
Fix CL direct conv layer test FP32 absolute tolerance issue
In CL/DirectConvolutionLayer tests, only relative tolerance was
being used, and this was causing issues for some small numbers.
Resolves: COMPMID-4755
Signed-off-by: Gunes Bayir <gunes.bayir@arm.com>
Change-Id: I976b70d393a0c196daf37047852eaf8894951e78
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6037
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_indirect.cpp')
0 files changed, 0 insertions, 0 deletions