js-examples | Code examples that accompany the MDN JavaScript | Script Programming library

I am trying to make a ChatBot using python (tensorflow/keras) for making, training and converting the neural network and then using it in my Angular app with tensorflow/tfjs. I was following the example found here: https://github.com/tensorflow/tfjs-examples/tree/master/translation but trying to add an embedding layer as well.

Creating the model:

I'm basically trying to achieve a kaleidoscopic effect with just one side, but I'm working with lots of Points, so I'd like that to happen in the shader. However if there's a Threejs trick that mirrors half of the texture or the Points object, that would be great. I tried to apply transformation matrices but I can't get it to work.

I found an old KaleidoShader that requires the usage of EffectComposer, but I'd like to implement it manually myself (without EffectComposer) and I'm struggling to do so. I'm using an FBO and I tried adding the code from that shader in both my simulation and render shaders but it's having no effect at all. Do I have to add yet another FBO texture or is it possibile to do those calculations in one of the existing shaders?

For visual reference https://ma-hub.imgix.net/wp-images/2019/01/23205110/premiere-pro-mirror-effect.jpg

I've spent so much time without getting to the bottom of this, hopefully someone can point me in the right direction.

Thanks

I am training a sequential tf.keras model which I want to convert to tfjs format consisting of a model.json file describing the layers and binary weight files to deploy it on a website for inference.

Two layers in my model are custom layers since there are no suitable layers predefined in tf.keras.layers to do the job. This is a mock version of what my model code looks like:

I have used the superimpose example to superimpose one pdf onto another. I am running into some trouble when trying to superimpose a page from a source file with a rotation of 90 degrees onto the destination page which has 0 rotation but is landscape (it's wider than it is tall).

When i first tried to stamp the content as in the example, it got the content rotated -90 as in the left example in the image below. When I tried to set the rotation of the destination page, the source content was placed in the right orientation but the destination page had turn (right example in the image below).

I've been trying to understand how attention mechanism works. Currently looking at tfjs-examples/date-conversion-attention example. I've found out that in the example the dot product alignment score (from Effective Approaches to Attention-based Neural Machine Translation) is being used.

So this expression is represented as

I'm trying to get the tensorflow.js example from the tensorflow.js-website running within my existing angular project. I basically copied the code from the website and integrated it into my component, but I'm receiving an error message.

Since I'm pretty new to angular I don't have any idea on how to fix the error.

I want to build a semi-complex neural network, so I'm not using tf.seqential().

I'm new using rails 5.1 with webpacker gem and came a across this issue while trying to configure my environment to use bpmn-js library. I installed the bpmn-js package with yarn but i still needed to add some required files from bpmn-js examples project to work properly in project/app/javascript/packs/application.js. The problem is that application.js uses fs module to create a new diagram as shown below:

project/app/javascript/packs/application.js

I'm trying to implement GradCam (https://arxiv.org/pdf/1610.02391.pdf) in tfjs, based on the following Keras Tutorial (http://www.hackevolve.com/where-cnn-is-looking-grad-cam/) and a simple image classification demo from tfjs, similar to (https://github.com/tensorflow/tfjs-examples/blob/master/webcam-transfer-learning/index.js) with a simple dense, fully-connected layer at the end.

However, I'm not able to retrieve the gradients needed for the gradcam computation. I tried different ways to retrieve gradients for the last sequential layer, but did not succeed, as types of tf.LayerVariable from the respective layer is not convertible to the respective type of tf.grads or tf.layerGrads.

Did anybody already succeeded to get the gradients from sequential layer to a tf.function like object?