distributions are the most commonly used, a lot of real-world data unfortunately is not normal. When faced with extremely skewed data, it’s tempting for us to utilize log transformations to normalize the distribution and stabilize the variance. I recently worked on a project analyzing the energy consumption of training AI models, using data from Epoch …
GPT-powered fashion assistant, I expected runway looks—not memory loss, hallucinations, or semantic déjà vu. But what unfolded became a lesson in how prompting really works—and why LLMs are more like wild animals than tools. This article builds on my previous article on TDS, where I introduced Glitter as a proof-of-concept GPT stylist. Here, I explore …
What My GPT Stylist Taught Me About Prompting Better Read More »
Today we’re releasing early access to Gemini 2.5 Pro Preview (I/O edition), an updated version of 2.5 Pro that has significantly improved capabilities for coding, especially building compelling interactive web apps. We were going to release this update at Google I/O in a couple weeks, but based on the overwhelming enthusiasm for this model, we …
ContentsEvery company I worked for until today, there it was: the resilient MS Excel. Excel was first released in 1985 and has remained strong until today. It has survived the rise of relational databases, the evolution of many programming languages, the Internet with its infinite number of online applications, and finally, it is also surviving …
Generating Data Dictionary for Excel Files Using OpenPyxl and AI Agents Read More »
We’ve seen developers doing amazing things with Gemini 2.5 Pro, so we decided to release an updated version a couple of weeks early to get into developers hands sooner. Today we’re excited to release Gemini 2.5 Pro Preview (I/O edition). This update features even stronger coding capabilities, for you to start building with before Google …
Gemini 2.5 Pro Preview: even better coding performance Read More »
posts, we explored Part I of the seminal book Reinforcement Learning by Sutton and Barto [1] (*). In that section, we delved into the three fundamental techniques underlying nearly every modern Reinforcement Learning (RL) algorithm: Dynamic Programming (DP), Monte Carlo methods (MC), and Temporal Difference Learning (TD). We not only discussed algorithms from each field …
Benchmarking Tabular Reinforcement Learning Algorithms Read More »
Research Published 31 August 2022 Authors Siqi Liu, Leonard Hasenclever, Steven Bohez, Guy Lever, Zhe Wang, S. M. Ali Eslami, Nicolas Heess Using human and animal motions to teach robots to dribble a ball, and simulated humanoid characters to carry boxes and play football Humanoid character learning to traverse an obstacle course through trial-and-error, which …
Contentsyou should read this Our Agenda:A Brief Note on Mathematical AbstractionThe “Normal” Norms: L1 and L2L¹ vs. L² Loss — Similarities and DifferencesL¹ Regularization (Lasso)L2 Regularization (Ridge)L¹ Loss in Generative Adversarial Networks (GANs)Why tiny differences matterGeneralizing Distance to LᵖThe L∞ NormConclusion you should read this As someone who did a Bachelors in Mathematics I was first introduced to L¹ and L² as a measure …
ContentsNew research drawing upon pragmatics and philosophy proposes ways to align conversational agents with human valuesInsights from pragmaticsDiscursive idealsPaths ahead New research drawing upon pragmatics and philosophy proposes ways to align conversational agents with human values Language is an essential human trait and the primary means by which we communicate information including thoughts, intentions, and …
In conversation with AI: building better language models Read More »
I’ve had the opportunity recently to be involved with a program called Skype a Scientist, which pairs scientists of various types (biologists, botanists, engineers, computer scientists, etc) with classrooms of kids to talk about our work and answer their questions. I’m pretty familiar with discussing AI and machine learning with adult audiences, but this is the …
