Monday:
9:30am: On my way to work, excited about this week as the project that we're working on shaping up really well.
1pm: met with the clients to understand and elaborate on the model that we're building. We're working on a model to predict the nuanced problem(home services) type based on the description that the customer is giving us.
6pm: time to hit the gym. happy with how the day progressed.
Tuesday:
11am: Time to prep the training and validation data for the model. Extracted historic home services jobs' description from the data lake.
4pm: Cleaning the data. The most important part of training any model. (What would I do without pandas).
6pm: Time to hit the gym. Happy how it's all shaping up.
Wednesday:
7:30am: Damn! out of the cereals. Need to pick up breakfast from office. Hope time permits.
10am: Exploring the new language model from google. google BERT. Impressive I must say. (missed the breakfast in case you are wondering)
4pm: I am convinced I'll use BERT for this model.
6pm: data prep is going well. the labelling is being done on amazon sagemaker. I'll call it a day. Need to go pick up cereals.
Thursday:
9:30 am: just heading to office(Yes! perks of living near office.)
11am: Was looking into using the pytorch-transformers for BERT.
6pm: entire day went in figuring out how best to use pytorch-transformers for BERT for the model. I am hopeful it performs well. Only time will tell.
Friyay! 10am: The labelling is done - Thanks to the business users. Need to pick up the files and convert them into the format that BERT understands.
3pm: Will be training the model entirely on AWS sagemaker. Will need to figure out how to provision resources on AWS. ah! so much learning.
5pm: I guess I'll call it a day. Great learning week. Looking forward to putting all the moving parts together next week. (wonder what to order in for the evening!)
