View Code? Open in Web Editor
NEW
A research project on running ML inferencing on KNative
Python 88.20%
Dockerfile 0.13%
Go 0.12%
Shell 0.09%
Jupyter Notebook 11.47%
anpr-knative's Introduction
๐ญ Iโm currently working on Kubernetes and Developer PaaS-es @VMware
๐ฑ Iโm currently learning Go and "real" development
๐ For more about me, and blogs, check out blah.cloud
๐ซ Contact me via Twitter
anpr-knative's People
Watchers
anpr-knative's Issues
This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.
Rate-Limited
These updates are currently rate-limited. Click on a checkbox below to force their creation now.
Open
These updates have all been created already. Click a checkbox below to force a retry/rebase of any.
Detected dependencies
dockerfile
label_analyser/Dockerfile
label_analyser/app/object_detection/dockerfiles/android/Dockerfile
label_analyser/app/object_detection/dockerfiles/tf1/Dockerfile
tensorflow/tensorflow 1.15.2-gpu-py3
label_analyser/app/object_detection/dockerfiles/tf2/Dockerfile
tensorflow/tensorflow 2.2.0-gpu
tensformation/Dockerfile
gomod
tensformation/go.mod
go 1.16
github.com/aws/aws-sdk-go v1.38.71
github.com/cloudevents/sdk-go/v2 v2.4.1
pip_requirements
label_analyser/requirements.txt
absl-py ==0.13.0
astor ==0.8.1
cached-property ==1.5.2
cachetools ==4.2.2
certifi ==2021.5.30
chardet ==4.0.0
click ==8.0.1
cloudevents ==1.2.0
deprecation ==2.1.0
Flask ==2.0.1
gast ==0.2.2
google-api-core ==1.30.0
google-api-python-client ==2.11.0
google-auth ==1.32.0
google-auth-httplib2 ==0.1.0
google-auth-oauthlib ==0.4.4
google-pasta ==0.2.0
googleapis-common-protos ==1.53.0
grpcio ==1.38.1
h5py ==3.3.0
httplib2 ==0.19.1
idna ==2.10
importlib-metadata ==4.6.0
imutils ==0.5.4
itsdangerous ==2.0.1
Jinja2 ==3.0.1
Keras-Applications ==1.0.8
Keras-Preprocessing ==1.1.2
Markdown ==3.3.4
MarkupSafe ==2.0.1
mpmath ==1.2.1
numpy ==1.21.0
nupy ==0.1.1
oauthlib ==3.1.1
opencv-python ==4.5.2.54
opt-einsum ==3.3.0
packaging ==20.9
protobuf ==3.17.3
pyasn1 ==0.4.8
pyasn1-modules ==0.2.8
pyparsing ==2.4.7
pytz ==2021.1
requests ==2.25.1
requests-oauthlib ==1.3.0
rsa ==4.7.2
six ==1.16.0
sympy ==1.8
tensorboard ==1.15.0
tensorflow ==1.15.0
tensorflow-estimator ==1.15.1
termcolor ==1.1.0
typing-extensions ==3.10.0.0
uritemplate ==3.0.1
urllib3 ==1.26.6
Werkzeug ==2.0.1
wrapt ==1.12.1
zipp ==3.4.1
I really do not like io.triggermesh.transformations.s3-tensorflow.response
I would like to see this replaced with io.triggermesh.transformations.tensformation.response
. However I want to wait for the demo to be recorded to avoid causing any havoc :)
TF Serving does not respond to SIGTERM, meaning when running in Docker/K8s it will timeout and be ended after the timeout by a SIGKILL:
Implement a wrapper script for the container entrypoint to take SIGTERM and automatically SIGKILL the child process as an interim step to having SIGTERM support added to TF Serving server upstream.
ref implementations:
Model warmup and pausing the container ready notification until the model has warmed up has the potential to increase performance by decreasing the latency per inference, and in particular on the first instance of the API's use per tf-inference container.
Need to generate model warmup data (which we can do with: https://github.com/mylesagray/tensorflow-anpr/tree/master/dataset_prep/artificial ), include in the anpr-serving container image and configure TF Serving to load the warmup data on start.
ref:
Currently we are using a set of numbers for batching in our tf-inference server that we have arrived at through trial and error.
There must be a more comprehensive way to look at these with a view to improving throughput and reliability.
refs: