experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Prefer android_arm64 for benchmark/eval scripts

Browse Source 
PiperOrigin-RevId: 303168086
Change-Id: Ia2d4eb1a348e71a72f7536af864200add4f02232
This commit is contained in:
Jared Duke 2020-03-26 12:18:04 -07:00 committed by TensorFlower Gardener
parent e410e94e52
commit 24f11dc4fc
2 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
tensorflow/lite/tools/accuracy/ilsvrc/README.md
View File

@ -99,7 +99,7 @@ python generate_validation_labels.py \
```
bazel build -c opt \
--config=android_arm \
--config=android_arm64 \
//tensorflow/lite/tools/accuracy/ilsvrc:imagenet_accuracy_eval
```

4
tensorflow/lite/tools/benchmark/README.md
View File

@ -112,7 +112,7 @@ and the following optional parameters:
```
bazel build -c opt \
--config=android_arm \
--config=android_arm64 \
tensorflow/lite/tools/benchmark:benchmark_model
```
@ -140,7 +140,7 @@ adb push mobilenet_quant_v1_224.tflite /data/local/tmp
That step is only needed when using the Hexagon delegate.
```
bazel build --config=android_arm \
bazel build --config=android_arm64 \
tensorflow/lite/experimental/delegates/hexagon/hexagon_nn:libhexagon_interface.so
adb push bazel-bin/tensorflow/lite/experimental/delegates/hexagon/hexagon_nn/libhexagon_interface.so /data/local/tmp
adb push libhexagon_nn_skel*.so /data/local/tmp