I have checked this errors come from the tfjs-backend-webgl
, as I am running some tensorflow code in the browser:
Assertion failed: gpgpu_util.bindVertexProgramAttributeStreams not fully successful
Assertion failed: gpgpu_util.bindVertexProgramAttributeStreams not fully successful.
Any ideas why would they appear in my tfjs code ? I am not even loading a neural network, just doing some tensor manipulation.
@chunduriv
Hi @Mah_Neh !
Either your code/graphics card/browser might be incompatible with WebGl. Have you checked with any other browser supporting WebGL. You can also downgrade or upgrade your TFJS version too. Just curious on your TFJS version .
Thank you.
1 Like
@Mah_Neh ,
Could you please provide standalone code to reproduce the problem? Please also include web browser and tfjs version information.
Thank you!
1 Like
Great, I think I manage to make a codepen that reproduces the error, it comes from the tf.fill
and if you agree I will add it to github, but maybe we can first make sure this could be a bug:
@chunduriv @ManuK001
BTW the juicy bit is the 3 liines of JS code and the error will appear in the console.
Same error seems to be reported by user in Github, but it is not scoped to a function
opened 07:37AM - 15 Apr 23 UTC
closed 01:49AM - 02 Aug 23 UTC
type:bug
comp:models
stat:awaiting response
stale
Hey! I just started with the ML YouTube tutorial and have an issue with the qna … model. I always get an empty array as an result. I tried it with several passages and questions. Besides that I also get:
`Assertion failed: gpgpu_util.bindVertexProgramAttributeStreams not fully successful.`
`Source-Map-Fehler: Error: request failed with status 404
Ressourcen-Adresse: https://cdn.jsdelivr.net/npm/@tensorflow/tfjs
Source-Map-Adresse: tf.min.js.map`
`WebGL warning: clientWaitSync: ClientWaitSync must return TIMEOUT_EXPIRED until control has returned to the user agent's main loop. (only warns once) `
I tried in in firefox and chrome. My Laptop is running Fedora.
This is from my HTML file:
`
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
<script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs"></script>
<script src="https://cdn.jsdelivr.net/npm/@tensorflow-models/qna"></script>
<script>
// Notice there is no 'import' statement. 'qna' and 'tf' is
// available on the index-page because of the script tag above.
const passage = "We belive cats are the real stars of YouTube";
const question = "What is important to YouTube?";
// Load the model.
qna.load().then((model) => {
console.log(model);
// Find the answers
model.findAnswers(question, passage).then((answers) => {
console.log("Answers: ", answers);
});
});
</script>`
@chunduriv @ManuK001 seems to be browser or platform specific. I use Ubuntu 20.04 and Firefox 111.0.1