Back to blog home

AI and MLOps Roundup: June 2023

LLMs have evolved, but what is ChatGPT actually thinking and how will this affect U.S. AI policy? Check out our roundup of the top AI and MLOps articles for June 2023!

Understanding Large Language Models

Want to deep dive into LLMs? Sebastian Raschka, PhD has compiled a great list of research papers covering different architectures and various techniques to improve transformer efficiency and fine-tune performance:

The evolutionary tree of modern LLMs

Unlocking Real-time Predictions with Shopify's Machine Learning Platform

Shopify customers require real-time predictions. Their ML team needed to upgrade their platform Merlin to be robust enough to handle this requirement and every function's use cases, while allowing low-latency and serving models at scale. Here's how they did it:

A high level overview of Merlin's user journey

What is ChatGPT thinking?

What is ChatGPT's reasoning for its responses? Understanding how LLMs think is central to develop responsible AI applications built on those models. We used a 'time travel' game to uncover some critical clues:

Flowchart diagram with GPT responses
ChatGPT's inconsistent responses

A Turning Point for U.S. AI Policy: Senate Explores Solutions

The US Senate met with AI leaders to understand how to shape AI policy. Marc Rotenberg, founder of the Center for AI and Digital Policy, describes the hearing's highlights and risks:

U.S. Senate building

Learning from deep learning: a case study of feature discovery and validation in pathology

Human learning from deep learning moves traditional ML from distilling existing knowledge to a tool for knowledge discovery. Google Health demonstrates how this can be used to improve cancer diagnosis and prognosis:

Coupling deep learning with interpretability methods provides an avenue for expanding the frontiers of scientific knowledge by learning from deep learning.

Reinforcement Learning from Human Feedback

Reinforcement learning from human feedback is critical to ChatGPT's success. But how and why does it work? Chip Huyen explains it all:

Where RLHF fits in

Stay up to date on everything AI and MLOps by signing up for our newsletter below.