Large Language Model-Based Solutions: How to Deliver Value with Cost-Effective Generative AI Applications by Shreyas Subramanian.
Shreyas holds a PhD in Aerospace Engineering from Purdue University and MS in Mechanical Engineering from Wright State University. He is the former Director of Research at Robust Analytics. Today he is Principal Data Scientist at Amazon Web Services.
Here is a good, very practical guide for those who seek to build and deploy cost-effective LLM-based solutions. From selecting a model, pre-and post-processing, prompt engineering, and fine tuning. Shreyas is certainly providing insights for optimizing inference and affordable architectures for typical applications. So today, generative AI value is found at the intersection of performance and cost. Howver organizations must optimize their infrastructure in order to reduce cloud costs.
Shreyas is certainly emphasizing the “biggest” model is not always the best. Model Selection and Foundation should be a wise, smaller approach provides developers to focus on domain-specific models. This requires less computational resources.