May 14Member-onlyWhat LayerNorm really does for Attention in Transformers2 things, not 1… Normalization via LayerNorm has been part and parcel of the Transformer architecture for some time. If you asked most AI practitioners why we have LayerNorm, the generic answer would be that we use LayerNorm to normalize the activations on the forward pass and gradients on the backward. But that default…Artificial Intelligence3 min readArtificial Intelligence3 min read
Jul 11, 2021Member-onlySculpted shoulders — the best exercise for the rear deltsBuilding from the V shaped physique concept, you need to also have depth from the side view. This is achieved by building the ‘rear delts’, and here’s the most efficient exercise for quickly doing that. The rear delts are often ignored in many workout routines — and they are hard…Body Image3 min readBody Image3 min read
Jul 5, 2021Member-onlyAchieving V shaped physique — the optimal shoulder exercise to add shoulder widthExercising your side delts can add impressive width, but one exercise gives you the best bang for the buck. Based on studies, women are most attracted to men with a V shaped physique, meaning the shoulders are wider than the waist, creating a V taper from the front view. Given…Bodybuilding3 min readBodybuilding3 min read
Jun 4, 2021Member-onlyA working, evidence based theory on the origins of Covid-19: hint, mice with human lung tissue, molecular dynamics, lab leak.Genetically modified mice with ‘humanized lungs’ were apparently created to test the infectiousness of coronaviruses in Summer, 2019 and not Spring 2020 as portrayed. Unlike other zoonotic jumps ala SARS, Covid-19 was surprisingly ‘pre-adapted’ to thrive and transmit in the human respiratory system from the start. Molecular modeling shows Covid-19…Coronavirus9 min readCoronavirus9 min read
Jun 1, 2021Member-onlyRapid fat loss, keep the muscle— has the optimal combo of diet and exercise been found?Combining intermittent fasting days with cardio = Double the fat loss, retain muscle mass, improve cholesterol… A study that compared the effects of intermittent fasting by itself, cardio by itself, or the combination of both may have found the optimal setup for rapid fat loss whilst maintaining muscle mass (and…Diet5 min readDiet5 min read
Jan 4, 2021Member-onlyBuilding Bigger Biceps — 4 science based tips that can maximize your results.Four tips from scientific studies point the way to maximizing your gains if you are interested in increasing your arm size as efficiently as possible! 1 — Train your arms first. While conventional workouts typically follow the larger muscles first, then smaller muscles, multiple studies have shown that muscles worked…Workout4 min readWorkout4 min read
Dec 27, 2020Member-onlySummary info on how the new mRNA Covid-19 vaccines (Pfizer, Moderna) work— no DNA reprogramming, no microchips:Short summary of how the US approved mRNA Covid vaccines work (Pfizer, Moderna) With my work in Covid-19 related AI, I frequently get asked about the various covid vaccines, both approved and pending. Thus, I’m writing this article to provide a useful summary of the current mRNA vaccines, and will…Covid-195 min readCovid-195 min read
Jan 5, 2020Member-onlyMeet AdaMod: a new deep learning optimizer with memoryAdaMod is a new deep learning optimizer that builds on Adam, but provides an automatic warmup heuristic and long term learning rate buffering. …Machine Learning5 min readMachine Learning5 min read
Dec 26, 2019Member-onlyMeet DiffGrad: New Deep Learning Optimizer that solves Adam’s ‘overshoot’ issueDiffGrad, a new optimizer introduced in the paper “diffGrad: An optimizer for CNN’s” by Dubey, et al, builds on the proven Adam optimizer by developing an adaptive ‘friction clamp’ and monitoring the local change in gradients in order to automatically lock in optimal parameter values that Adam can skip over.Machine Learning5 min readMachine Learning5 min read
Dec 16, 2019Member-onlyReducing your labeled data requirements (2–5x) for Deep Learning: Deep Mind’s new “Contrastive Predictive Coding 2.0”Current Deep Learning for vision, audio, etc. requires vast amounts of human labeled data, with many examples of each category, to properly train a classifier to acceptable accuracy. By contrast, humans only need to see a few examples of a class to begin properly and accurately recognizing and classifying future…Machine Learning5 min readMachine Learning5 min read
