To get help from the community, we encourage using Stack Overflow and the tensorflow.js tag.
0.15.3
Chrome Version 72.0.3626.119
Simple include the tensorflow in a html, and it always tries to register with webgl and fails. I confirm my chrome does support webgl. I even tried to set backend to cpu, and this error still appears (in the console).

Just a simple html as below with tfjs loaded in script tag. You can see I am using latest as of today.
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta http-equiv="X-UA-Compatible" content="ie=edge">
<script src="https://cdn.jsdelivr.net/npm/@tensorflow/[email protected]/dist/tf.min.js"></script>
<title>Testing Tfjs</title>
</head>
<body>
<h2>Testing Tfjs</h2>
</body>
</html>
You might need to just restart Chrome. Chrome sometimes can get in a bad state w.r.t WebGL.
Also, does this site show WebGL is enabled? http://webglreport.com/
Oh by the way, this is something that should be posted on stackoverflow.com.
yes, I did restart Chrome and same issue. Yes, I checked those links and they say my browser is compatable with webgl. Can you please explain why this is to go to SO?
Even when the browser does support webgl, the tf is failing, and this happens with just loading the tfjs script without doing anything at all. So clearly tf has a problem correctly evaluating webgl compatability. I have seen this issue multiple times posted about tfjs, that is why tried latest one thinking it might be fixed, but the same exact problem pops up.
If you follow the stack trace can you find where the exception is thrown and see what's the issue? This error happens when we catch any exceptions when registering the backend.
Also, what hardware are you on?
I got the same error on an Xeon E2 1200 CPU with Intel graphics.
The error message is not very explicit on the version or the webgl extension it needs as other webgl examples are running (three.js).
Can you post what you see from http://webglreport.com/?
These are the report from the PC having the Xeon and where tf.js doesn't work accelerated.


Here it works without any seen problems.


Automatically closing due to lack of recent activity. Please update the issue when new information becomes available, and we will reopen the issue. Thanks!
I experienced this same behavior when importing tfjs-core in a static page that rendered a WebGL canvas on a Linux Mint 19.1 device running recent Chromium.
Most helpful comment
I experienced this same behavior when importing tfjs-core in a static page that rendered a WebGL canvas on a Linux Mint 19.1 device running recent Chromium.