Mousius commented on a change in pull request #9979:
URL: https://github.com/apache/tvm/pull/9979#discussion_r787912180
##########
File path: apps/microtvm/ethosu/README.md
##########
@@ -72,15 +73,15 @@ the locations for these can be specified as arguments to
run_demo.sh, for exampl
```
This will:
-- Download a quantized mobilenet v1 model
-- Use tvmc to compile the model for Cortex(R)-M55 CPU and Ethos(TM)-U55 NPU
-- Download an image of a kitten to run the model on
+- Download a quantized (int8) mobilenet v2 model
+- Use tvmc to compile the model for Cortex(R)-M55 CPU, Ethos(TM)-U55 NPU and
CMSIS-NN
+- Download an image of a penguig to run the model on
Review comment:
I don't think this hybrid is the one we're testing for.
##########
File path: apps/microtvm/ethosu/Makefile
##########
@@ -57,6 +59,7 @@ CODEGEN_SRCS = $(wildcard $(abspath
$(BUILD_DIR))/codegen/host/src/*.c)
CODEGEN_OBJS = $(subst .c,.o,$(CODEGEN_SRCS))
CMSIS_STARTUP_SRCS = $(wildcard ${CMSIS_PATH}/Device/ARM/${ARM_CPU}/Source/*.c)
UART_SRCS = $(wildcard ${CORSTONE_300_PATH}/*.c)
+CMSIS_NN_LIBS = $(wildcard ${CMSIS_PATH}/CMSIS/NN/build/Source/*/*.a)
Review comment:
For the purposes of the demo we should reference the `.c` files and not
rely on these being built externally.
##########
File path: apps/microtvm/ethosu/run_demo.sh
##########
@@ -123,32 +123,32 @@ script_dir="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &>
/dev/null && pwd )"
mkdir -p build
cd build
-# Get mobilenet_v1 tflite model
-mobilenet_url='https://storage.googleapis.com/download.tensorflow.org/models/mobilenet_v1_2018_08_02/mobilenet_v1_1.0_224_quant.tgz'
-curl --retry 64 -sSL ${mobilenet_url} | gunzip | tar -xvf -
./mobilenet_v1_1.0_224_quant.tflite
+# Get mobilenet_v2 tflite model
+mobilenet_url='https://github.com/ARM-software/ML-zoo/raw/master/models/image_classification/mobilenet_v2_1.0_224/tflite_int8/mobilenet_v2_1.0_224_INT8.tflite'
Review comment:
We recently had issues with referencing `master`
(https://github.com/apache/tvm/pull/9887) and the model changing, can you pick
a SHA1 that guarantees we'll always get the same model?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
