Tensorflow Model Experiment Tracking using

[@AshutoshTripathi_AI]

MLFlow complete demo playlist: kzbin.info/aero/PLwFaZuSL_mfrCjOdA6I--wC13Hj7eNRYU

[@pinkfloyd2642]

Great job, Ashutosh! I am kind of unhappy with the KZbin algo for not promoting your videos much. You definitely deserve more views and subscribers. The MLFlow playlist is very good.

[@AshutoshTripathi_AI]

Thank you so much for your kind words. You guys keep watching and supporting that day will also come 🙏

[@sanketsaurabh5881]

great video, after a break

[@AshutoshTripathi_AI]

Yeah, sometime lots of work come in between so do not get time, however I always try to create useful live demo videos on topics which i had faced in my day to day work and thinking it might be helpful to others. btw than you very much for watching my videos.

[@sankalpchenna6046]

Great content, if possible can you make a video where we can store artifacts in an S3 bucket(minio)

[@ankitmaheshwari7310]

What if we have custom metric? How to add this in artifect?

[@PiyushRaj-on7en]

I trained tflite model for android and want to experiment with mlflow on google colab...but autolog is not working and also mlflow.tensorflow.log_model is not working..my artifacts are not sav in the mlflow ui..I am using 2.8 version of tensorflow.....please give a solution for this

[@s.seducation3888]

Can you show how to serve and create an inference for prediction for Image classification in that serving end point.

[@AshutoshTripathi_AI]

Ok, will create and upload soon. Thanks for watching.

[@siddharthtyagi1254]

sir is there are something else to track the nlp model params.

[@AshutoshTripathi_AI]

You can refer to this link until I make a video on explaining it. 😀 github.com/mlflow/mlflow/blob/master/examples/spacy/train.py

[@siddharthtyagi1254]

​@@AshutoshTripathi_AI okay

[@pedromartinezbarron4720]

The mlflow experiment schema/tracking presented should also work if im loading data differently (not specific from tf datasets, rather custom dataset)?

[@AshutoshTripathi_AI]

Yes of course it will work.

[@Sam-nn3en]

Please make a video on Argo Workflows. It will be interesting to see that since its purely YAML based, how and where in ML lifecycle can it be used. Also, if possible please compare pros and cons of Argo Workflow with kubeflow