Hello :) Today is Day 164!
A quick summary of today:
- started Module 4: deployment from the MLOps zoomcamp
- re-creating a baseline OD demand matrix model - ConvLSTM
Frstly, about module 4: deployment
Code from today is on my github. The first part was about simple model packaging with pickle and serving the model through a flask api
Second part was about using mlflow and saving the ran experiments to S3 on AWS The code is pretty much like the one in the pic above, but with the below added above it:
RUN_ID = ""
AWS_S3_BUCKET = ""
logged_model = f's3://{AWS_S3_BUCKET}/1/{RUN_ID}/artifacts/model'
model = mlflow.pyfunc.load_model(logged_model)
And this is when I realised I am getting charged by AWS. It is only $9, but I cannot continue using it. I thought I am under some ‘free tier’ but nope… I started deleting all EC2, RDS, Load balancers, Users, everything I could see. But because I am not completely sure, I wanted to use some cloud-nuke tool - and faced multiple errors with it. So finally, I contacted AWS support and they let me know what else I need to delete/cancel so that I do not get charged more. For any AWS/cloud parts, I might skip for now, at least until I figure out how can I get free credits for GCP (or if there are for AWS).
The next bit is about streaming models and using AWS Kinesis and Lambda, but I will just watch along on this one haha.
Secondly, about what I read and did in the lab today
Mostly, I was trying to recreate baseline models. Today I had my eye on ConvLSTM
I only took this picture - showing predicted and ground truth OD matrix (and the diff). It took some time to train because my google colab good GPU compute credits were exhausted :/ Tomorrow I might bring my laptop to the lab so I can use it to train and play around with hyper params.
That is all for today!
See you tomorrow :)