ML

Most of us got into data science because it’s exciting (if chaotic) and there’s a constant stream of new ideas, which is thrilling (if intimidating). But if learning is all about keeping up, why can’t our models do it?

Real-time ML models continually learn on new data as soon as it arrives, so they’re less susceptible to concept drift and data drift. Read on to learn how to use River and Vowpal Wabbit to build real-time models in Python.

Thinking about using a large language model (LLM) at your organization? Check out this tutorial to see how to bootstrap an MVP using an open source pre-trained model from HuggingFace and a free Ensign account.

A few weeks ago, I’d never used an event stream before. Read on to hear how I built my first event-driven data science app — the biggest challenges, my lessons learned, and a couple of key takeaways!

If you can’t deploy your models, you might feel frustrated, but you aren’t alone — only 1 in 10 data science projects ever gets deployed. The fix? We need to shift our mental model towards “thinking in events.”

Profitable businesses rely on their own data to grow, make smart choices, and delight their customers. Easier said than done! While a batched approach can provided limited success, real-time data analytics often give better ROI.

The release of artificially intelligent tools for public use has spurred a heated debate in the past few years. While AI art was a lark for some and an opportunity to experience gender euphoria for others, to artists it represented stolen …

With almost 30 billion connected devices, this isn’t Al Gore’s internet anymore. Consider this: According to Cisco’s Annual Internet Report, there will be 29.3 billion internet-connected devices in 2023, or more than three for every human. …

If you want to build a robust machine learning model, the most important ingredient is data – but keep in mind that tuning your model will rely on devising a systematic way to store and query that data! In this post, we explore a …

In this post, we’ll consider how it is that models trained on massive datasets using millions of parameters can be both “low bias” and also very biased, and begin to think through what we in the ML community might be able …