Effective MLOps: Model Development using W&B

This repository details work done on:

• Building a segmentation model prototype end-to-end (link).
• Optimising the performance of the baseline model (link).
• Best practices for model evaluation (link).

W&B tools used:

• Tables
• Artifacts
• Experiments
• Reports
• Model registry

Dataset used:

• BDD1K dataset to be used for semantic segmentation.

Further W&B resources:

• https://docs.wandb.ai/tutorials

Repository setup

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
wandb login