This is the TensorFlow agent for MLModelScope, an open-source framework and hardware agnostic, extensible and customizable platform for evaluating and profiling ML models across datasets / frameworks / systems, and within AI application pipelines.
Check out MLModelScope to learn more and to contribute.
We first discuss a bare minimum tensorflow-agent installation without the tracing and profiling capabilities. To make this work, you will need to have the following system libraries preinstalled in your system.
- The CUDA library (required)
- The CUPTI library (required)
- The Tensorflow library (required)
- The libjpeg-turbo library (optional, but preferred)
Please refer to Nvidia CUDA library installation on this. Find the location of your local CUDA installation, which is typically at /usr/local/cuda/, and setup the path to the libcublas.so library. Place the following in either your ~/.bashrc or ~/.zshrc file:
export LIBRARY_PATH=$LIBRARY_PATH:/usr/local/cuda/lib64
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64
Please refer to Nvidia CUPTI library installation on this. Find the location of your local CUPTI installation, which is typically at /usr/local/cuda/extras/CUPTI, and setup the path to the libcupti.so library. Place the following in either your ~/.bashrc or ~/.zshrc file:
export LIBRARY_PATH=$LIBRARY_PATH:/usr/local/cuda/extras/CUPTI/lib64
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/extras/CUPTI/lib64
The TensorFlow C library is required for the TensorFlow Go package. If you want to use TensorFlow Docker Images (e.g. NVIDIA GPU CLOUD (NGC)) instead, skip this step for now and refer to our later section on this.
You can download pre-built TensorFlow C library from Install TensorFlow for C.
Extract the downloaded archive to /opt/tensorflow/.
tar -C /opt/tensorflow -xzf (downloaded file)
Configure the linker environmental variables since the TensorFlow C library is extracted to a non-system directory. Place the following in either your ~/.bashrc or ~/.zshrc file
Linux
export LIBRARY_PATH=$LIBRARY_PATH:/opt/tensorflow/lib
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/tensorflow/lib
macOS
export LIBRARY_PATH=$LIBRARY_PATH:/opt/tensorflow/lib
export DYLD_LIBRARY_PATH=$DYLD_LIBRARY_PATH:/opt/tensorflow/lib
You can test the installed TensorFlow C library using an example C program.
To build the TensorFlow C library from source, refer to TensorFlow in Go .
libjpeg-turbo is a JPEG image codec that uses SIMD instructions (MMX, SSE2, AVX2, NEON, AltiVec) to accelerate baseline JPEG compression and decompression. It outperforms libjpeg by a significant amount.
You need libjpeg installed.
sudo apt-get install libjpeg-dev
The default is to use libjpeg-turbo, to opt-out, use build tag nolibjpeg.
To install libjpeg-turbo, refer to libjpeg-turbo.
Linux
export TURBO_VER=2.0.2
cd /tmp
wget https://cfhcable.dl.sourceforge.net/project/libjpeg-turbo/${TURBO_VER}/libjpeg-turbo-official_${TURBO_VER}_amd64.deb
sudo dpkg -i libjpeg-turbo-official_${TURBO_VER}_amd64.deb
macOS
brew install jpeg-turbo
Since we use go for MLModelScope development, it's required to have go installed in your system before proceed.
Please follow Installing Go Compiler to have go installed.
Download and install the MLModelScope TensorFlow Agent by running the following command in any location, assuming you have installed go following the above instruction.
go get -v github.com/c3sr/tensorflow
You can then install the dependency packages through go get.
cd $GOPATH/src/github.com/c3sr/tensorflow
go get -u -v ./...
An alternative to install the dependency packages is to use Dep.
dep ensure -v
This installs the dependency in vendor/.
The CGO interface passes go pointers to the C API. There is an error in the CGO runtime. We can disable the error by placing
export GODEBUG=cgocheck=0
in your ~/.bashrc or ~/.zshrc file and then run either source ~/.bashrc or source ~/.zshrc
Build the TensorFlow agent with GPU enabled
cd $GOPATH/src/github.com/c3sr/tensorflow/tensorflow-agent
go build
Build the TensorFlow agent without GPU or libjpeg-turbo
cd $GOPATH/src/github.com/c3sr/tensorflow/tensorflow-agent
go build -tags="nogpu nolibjpeg"
If everything is successful, you should have an executable tensorflow-agent binary in the current directory.
To run the agent, you need to setup the correct configuration file for the agent. Some of the information may not make perfect sense for all testing scenarios, but they are required and will be needed for later stage testing. Some of the port numbers as specified below can be changed depending on your later setup for those service.
So let's just set them up as is, and worry about the detailed configuration parameter values later.
You must have a carml config file called .carml_config.yml under your home directory. An example config file carml_config.yml.example is in github.com/c3sr/MLModelScope . You can move it to ~/.carml_config.yml.
The following configuration file can be placed in $HOME/.carml_config.yml or can be specified via the --config="path" option.
app:
name: carml
debug: true
verbose: true
tempdir: ~/data/carml
registry:
provider: consul
endpoints:
- localhost:8500
timeout: 20s
serializer: jsonpb
database:
provider: mongodb
endpoints:
- localhost
tracer:
enabled: true
provider: jaeger
endpoints:
- localhost:9411
level: FULL_TRACE
logger:
hooks:
- syslogWith the configuration and the above bare minimumn installation, you should be ready to test the installation and see how things work.
Here are a few examples. First, make sure we are in the right location
cd $GOPATH/src/github.com/c3sr/tensorflow/tensorflow-agent
To see a list of help
./tensorflow-agent -h
To see a list of models that we can run with this agent
./tensorflow-agent info models
To run an inference using the default DNN model mobilenet_v1_1.0_224 with a default input image.
./tensorflow-agent predict urls --profile=false --publish=false
The above --profile=false --publish=false command parameters tell the agent that we do not want to use profiling capability and publish the results, as we haven't installed the MongoDB database to store profiling data and the tracer service to accept tracing information.
We now discuss how to install a few external services that make the agent fully useful in terms of collecting tracing and profiling data.
MLModelScope relies on a few external services. These services provide tracing and database servers.
These services can be installed and enabled in different ways. We discuss how we use docker below to show how this can be done. You can also not use docker but install those services from either binaries or source codes directly.
Refer to Install Docker.
On Ubuntu, an easy way is using
curl -fsSL get.docker.com -o get-docker.sh | sudo sh
sudo usermod -aG docker $USER
On macOS, intsall Docker Destop
This service is required.
- On x86 (e.g. intel) machines, start jaeger by
docker run -d -e COLLECTOR_ZIPKIN_HTTP_PORT=9411 -p5775:5775/udp -p6831:6831/udp -p6832:6832/udp \
-p5778:5778 -p16686:16686 -p14268:14268 -p9411:9411 jaegertracing/all-in-one:latest
- On ppc64le (e.g. minsky) machines, start jaeger machine by
docker run -d -e COLLECTOR_ZIPKIN_HTTP_PORT=9411 -p5775:5775/udp -p6831:6831/udp -p6832:6832/udp \
-p5778:5778 -p16686:16686 -p14268:14268 -p9411:9411 carml/jaeger:ppc64le-latest
The trace server runs on http:https://localhost:16686
This service is not required if not using database to publish evaluation results.
- On x86 (e.g. intel) machines, start mongodb by
docker run -p 27017:27017 --restart always -d mongo:3.0
You can also mount the database volume to a local directory using
docker run -p 27017:27017 --restart always -d -v $HOME/data/carml/mongo:/data/db mongo:3.0
You must have a carml config file called .carml_config.yml under your home directory. An example config file ~/.carml_config.yml is already discussed above. Please update the port numbers for the above external services accordingly if you decide to choose a different ports above.
The testing steps are very similar to those testing we discussed above, except that you can now safely use both the profiling and publishing services.
Run ./tensorflow-agent -h to list the available commands.
Run ./tensorflow-agent predict to evaluate a model. This runs the default evuation.
./tensorflow-agent predict -h shows the available flags you can set.
An example run is
./tensorflow-agent predict general --trace_level=FRAMEWORK_TRACE --model_name=Inception_v3
We have pre-built docker images on Dockerhub. The images are c3sr/tensorflow-agent:amd64-cpu-latest, c3sr/tensorflow-agent:amd64-gpu-latest. The entrypoint is set as tensorflow-agent thus these images act similar as the command line above.
An example run is
docker run --gpus=all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 --privileged=true \
--network host \
-v ~/.carml_config.yml:/root/.carml_config.yml \
-v ~/results:/go/src/github.com/c3sr/tensorflow/results \
c3sr/tensorflow-agent:amd64-gpu-latest predict urls --trace_level=FRAMEWORK_TRACE --model_name=Inception_v3
NOTE: The SHMEM allocation limit is set to the default of 64MB. This may be insufficient for TensorFlow. NVIDIA recommends the use of the following flags:
--shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 ...
NOTE: To run with GPU, you need to meet following requirements:
- Docker >= 19.03 with nvidia-container-toolkit (otherwise need to use nvidia-docker)
- CUDA >= 10.1
- NVIDIA Driver >= 418.39
Build TensorFlow 1.14.0 with the following scripts.
go get -d github.com/tensorflow/tensorflow/tensorflow/go
cd ${GOPATH}/src/github.com/tensorflow/tensorflow
git fetch --all
git checkout v1.14.0
./configureConfigure the build and then run
bazel build -c opt //tensorflow:libtensorflow.so
cp ${GOPATH}/src/github.com/tensorflow/tensorflow/bazel-bin/tensorflow/libtensorflow.so /opt/tensorflow/lib
Need to put the directory that contains libtensorflow_framework.so and libtensorflow.so into $PATH.
For TensorFlow compilation, here are the recommended tensorflow-configure settings:
export CC_OPT_FLAGS="-mcpu=power8 -mtune=power8"
export GCC_HOST_COMPILER_PATH=/usr/bin/gcc
ANACONDA_HOME=$(conda info --json | python -c "import sys, json; print json.load(sys.stdin)['default_prefix']")
export PYTHON_BIN_PATH=$ANACONDA_HOME/bin/python
export PYTHON_LIB_PATH=$ANACONDA_HOME/lib/python2.7/site-packages
export USE_DEFAULT_PYTHON_LIB_PATH=0
export TF_NEED_CUDA=1
export TF_CUDA_VERSION=9.0
export CUDA_TOOLKIT_PATH=/usr/local/cuda-9.0
export TF_CUDA_COMPUTE_CAPABILITIES=3.5,3.7,5.2,6.0,7.0
export CUDNN_INSTALL_PATH=/usr/local/cuda-9.0
export TF_CUDNN_VERSION=7
export TF_NEED_GCP=1
export TF_NEED_OPENCL=0
export TF_NEED_HDFS=1
export TF_NEED_JEMALLOC=1
export TF_ENABLE_XLA=1
export TF_CUDA_CLANG=0
export TF_NEED_MKL=0
export TF_NEED_MPI=0
export TF_NEED_VERBS=0
export TF_NEED_GDR=0
export TF_NEED_S3=0