`modelAssetPath` method of loading in `LlmInference.createFromOptions` results in inefficient loading and crashes

[RandomGamingDev]

MediaPipe Solution (you are using)

@mediapipe/tasks-genai v0.10.25

Programming language

Javascript/Typescript

Are you willing to contribute it

Yes

Describe the feature and the current behaviour/state

Loading with

Will this change the current API? How?

No.

Who will benefit with this feature?

Anyone who wants to load a model since it benefits performance, latency, the number of models loaded, etc.

Please specify the use cases for this feature

Better loading models.

Any Other info

Relevant PR from mediapipe-samples

[RandomGamingDev]

@kuaashish I noticed that you were assigned to the issue, but I already created a fix. Mind if I create the PR?

[HarishPermal]

Hi @RandomGamingDev,

Thanks for filing this and for the clear context, feel free to open a PR here referencing this issue and that sample PR.