I am about to start a PoC in my company for some use cases that are identified by the business team members.
My question is how do I go about creating a hypothesis for the model selection step? Being a microsoft shop, there is a assumption that we will use openAI API’s , however after doing the week 1 of the course , I realized that selecting the model is a step . I didn’t see it elaborated on the course, so will appreciate any advise on criteria’s to follow for model selection
Hello @Rachana_Dikshit ,
To create a hypothesis for model selection in your PoC, consider the problem type, data availability, performance metrics, model complexity, computational resources, pretrained models, documentation, and cost. Determine if it’s a supervised, unsupervised, or reinforcement learning problem. Assess data availability, size, and labeling. Define performance metrics aligned with your objectives. Evaluate model complexity, interpretability, and computational requirements. Check for existing pretrained models or transfer learning possibilities. Explore documentation and community support. Consider cost and licensing implications. Iterate and experiment to find the optimal model for your specific use cases.