Replies: 1 comment 2 replies
-
|
wasm modules used by wasm backend cannot be packed like that, they need to be shipped and loded using http. |
Beta Was this translation helpful? Give feedback.
-
|
I've not manually used any wasm modules for the human library. |
Beta Was this translation helpful? Give feedback.
-
|
After further debugging, I found the issue could be related to TensorFlow.js WASM multithreading support. By manually overriding the TensorFlow environment flags on the page where the error occurred using the following snippet: the wasm log changed from Interestingly, when I imported the source code directly via file path (import ... from '../../abc'), the WASM log still showed Here's a link to a reproducible repo - https://github.com/Smruti-AT/test/tree/master Any idea about this? |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
I've created a JavaScript SDK (ABC) that uses the human library and tensorflow/tfjs for ML operations. I bundle the SDK using Webpack v5.94.
Everything works fine when I use the source code directly in my Next.js app using
import { ABC } from "../../abc.js". However, when I install the SDK from npm (i.e., the bundled output), and then useimport { ABC } from "@ABC/abc"I encounter the following error during the ML operationsThe output type in the webpack config file is "module" and my package type is also "module" for both the SDK and my NextJS app.
Any ideas on what might be wrong with this setup?
Thanks in advance!
Beta Was this translation helpful? Give feedback.
All reactions