'm posting here the link of a colab notebook with the test for the new GPT-2 model released by openAI https://colab.research.google.com/drive/19Yha3-EtYxumOInqaZkTm1c7nCD7ZlYk#scrollTo=cMes5yRQuXs4
GPT2 is a model developed by OpenAI for NLP. The model is composed by a transformer architecture as shown in the picture below. Luckily for us the dudes at Huggingface have releases a pretrained version and implementation of this model ... and many other type of transformer models. https://huggingface.co/
Recently I moved to Uruguay, Montevideo. Here I found an active community of developers, yesterday I had a chance of giving a talk at the AngularMVD meetup. Even though it has been ages from the last time I developed an Angular app the talk was about how to use tensorflow.
Recently I've been using Parcel (https://parceljs.org/) for some small web apps I'm developing in my free time. Parcel is a modern javascript bundler that is used to package a JS app into a single, or multiple files. We could argue that the state of
Send Your Name to Mars. NASA's Mars 2020 Rover is heading to the red planet. Submit your name by Sept. 30, 2019, and fly along! Your email is used only to give you “Frequent Flyer” points and to allow you to send your name on future Mars missions.
Transfer learning is a popular approach used in machine learning where a model trained on a task is re-purpose a model designed for a different task. Therefore the definition of Transfer Learning is the following. Given a source domain D(s) and learning task T(s), a target domain D(
Here is a simple tutorial on how to use the mobilenet model released by google with tensorflow-js. This tutorial requires node v >10 and the parcel bundler. To install node use nvm https://github.com/nvm-sh/nvm and to install parcel just run npm -i g parcel-bundler step 1
Tensorflow is the defacto standard for building advanced neural networks of any kind. Natively tensorflow is a python library and gives to the user a framework that simplifies operation with tensors. A tensor is a mathematical object analogous to but more general than a vector, represented by an array of
Playing with models released for tensorflow js. Posenet is a neural network that allows the estimation of a human pose from an image. For more technical information about how it actually works refer to this blog post https://medium.com/tensorflow/real-time-human-pose-estimation-in-the-browser-with-tensorflow-js-7dd0bc881cd5 import * as posenet from "@tensorflow-models/posenet"
I needed a cool way to visualize a force directed graph with many datapoints, so I thought that webGL was the way to go as it should be able to handle the heavy load better than d3 SVG. The library I've used for this small proof of concept