# size

## Sample size calculation

If you’re going to run a test on rabbits, you have to decide how many rabbits you’ll use. This is…

## GPT-3, a giant step for Deep Learning and NLP?

“The diversity of tasks the model is able to perform in a zero-shot setting suggests that high-capacity models trained to…

## How to estimate ODE solver error

This post brings together several themes I’ve been writing about lately: caching function evaluations, error estimation, and Runge-Kutta methods. A…

## Follow & Learn: Experiment Size With Python

at 80%. The plot below shows the distributions under null and test hypotheses. Analytical ApproachLet’s take the plot above and…

## Master Your Hypothesis Test: A tutorial on Power, Bootstrapping, Sample Selection, and Outcome Analysis.

This is a very nice sanity check, and the formula also gives us a P-Value. 6. Evaluating our P-ValueWhen we…

## Simplicity in the complexity

Let’s break down the term into “neuron” and “network”. To interpret “neuron”, we can think of it as a black…

## Scaling Transformer-XL to 128 GPUs

Scaling Transformer-XL to 128 GPUsBen MannBlockedUnblockFollowFollowingMay 9Ben Mann, Yaroslav Bulatov, Darius LamTL;DR: we made Transformer-XL train efficiently on 128 GPUs on…

## Sample size and correlation

As a rule of thumb, statisticians consider an r-value of 0. 7 to be strong. Let’s take a look at…

## Big O Notation (using Ruby)

Big O Notation (using Ruby)Daniel KoBlockedUnblockFollowFollowingApr 11Big O Notation allows us to calculate the worst possible runtime of an algorithm, or…

## How Do You Know You Have Enough Training Data?

What Happens in the Case of Deep Learning?Figure 1. Figure 1 shows how the performance of machine learning algorithms changes…

## Scala: Useful User Information + Why you should consider the Language

") } val arr = Array. fill(2097152)(0) tenMillionSearches(arr, 2097152)}Running the program above, we obtain the following results:Scala vs.  C++Now we…

## Predicting the performance of deep learning models

Predicting the performance of deep learning modelsPower-law scaling explains how a model’s performance will change as we feed it more dataArchy de…

## Better Heatmaps and Correlation Matrix Plots in Python

Finding the highest negative and positive correlations mean finding the strongest red and green. To do that I need to…

That number is sure to go up, if not significantly. 3. 9MB is pretty huge when Medium’s homepage, for example…

## Thunderstruck: Disaster CNN visualization of AC power lines

What during all the training made it look there for being the most significant area?That being said, this prediction should…

## Go memory ballast: How I learned to stop worrying and love the heap

Go memory ballast: How I learned to stop worrying and love the heapRoss EngersBlockedUnblockFollowFollowingApr 10I’m a big fan of small code…

## Deep learning to identify Malaria cells using CNN on Kaggle

Deep learning to identify Malaria cells using CNN on KaggleKaran BhanotBlockedUnblockFollowFollowingApr 5Photo by Kendal James on UnsplashDeep learning has vast ranging applications…

## Predicting Stock Price with LSTM

", df_ge. isna(). sum())Normalizing the dataThe data is not normalized and the range for each column varies, especially Volume. Normalizing data…

## Election Poll Simulation, Margin of Error and Central Limit Theorem with Python

Election Poll Simulation, Margin of Error and Central Limit Theorem with PythonWaldecir FariaBlockedUnblockFollowFollowingMar 14While learning about the Central Limit Theorem (CLT)…

## MezzFS — Mounting object storage in Netflix’s media processing platform

MezzFS — Mounting object storage in Netflix’s media processing platformNetflix Technology BlogBlockedUnblockFollowFollowingMar 6By Barak Alon (on behalf of Netflix’s Media Cloud Engineering…

## Problem Analysis of Code Jam to I/O for Women’19

(Think of some greedy approach)3. Sleep WalkingProblem Description -We have a 2-D grid with infinite number of unit cells. A person…

## Data Science in the Trenches: Living w/ Small n

What can we do?”Well, there’s a couple of things that you can consider, all of which involve different kinds of…

## A Significant Answer to that Statistical Question

You will need buy-in to talk to more people now, so build up your stakeholders’ appetite for discovery by presenting…