Test Env:
Chromium Version: nightly build 75.0.3739.0 (3acec85)
Platform: macOS/Linux/Andriod/Windows
Expected Result:
The probability of all models in benchmark should be match with the example.
Actual Result:
The probability of some models in benchmark does not match with the example when run Inception v3/Inception v4/Inception Resnet v2(TFLite) (except for Inception Resnet v2(TFLite) on MPS):
Test Env: Chromium Version: nightly build 75.0.3739.0 (3acec85) Platform: macOS/Linux/Andriod/Windows
Expected Result: The probability of all models in benchmark should be match with the example.
Actual Result: The probability of some models in benchmark does not match with the example when run Inception v3/Inception v4/Inception Resnet v2(TFLite) (except for Inception Resnet v2(TFLite) on MPS):
For example: Inception v3 clDNN on Windows: