Charlie Ruan

Closing this issue for now due to staleness. Try smaller models (e.g. Phi3.5, Qwen2) and see if the same problem is observed. Feel free to re-open new issues as well.

What device and browser are you using? We'd recommend Chrome Canary; Chrome 121 would work as well after its release. Could you try if https://webgpureport.org/ works on your Android first?

Thanks for reporting the issue. Device lost is usually due to not having enough memory; perhaps you can try smaller models like Gemma-2B. You can check out the field `vram_required_MB`...

Hi @skyser2003! Apologies for the inconvenience. This should be fixed now via https://github.com/mlc-ai/mlc-llm/pull/2187. Try the npm 0.2.34 with the newly compiled model.

Thank you for the work! Regarding the big code models, I think it should be fixed by https://github.com/mlc-ai/mlc-llm/pull/1515

For WizardLM-7B-V1.0 and WizardLM-30B-V1.0, their weights on HF [are weight delta](https://github.com/mlc-ai/mlc-llm/pull/489); and I think they are somewhat obsolete already (they are pre-llama2). We can probably just support [WizardLM-13B-V1.2](https://huggingface.co/WizardLM/WizardLM-13B-V1.2) and [WizardLM-70B-V1.0](https://huggingface.co/WizardLM/WizardLM-70B-V1.0)...

@junrushao Ahh yes! WizardCoder is fixed by setting context window size to smaller -- I think it is working fine! I added a WizardLM and two smaller WizardCoder Python to...

Hi @bulutthecat @qc903113684 apologies for the inconvenience. Could you check whether https://github.com/mlc-ai/mlc-llm/pull/1955 was included when you ran into this issue? Or perhaps try again with the latest package? I suspect...

Hi @qc903113684, #1850 is superseded by #1822, which was merged 3 weeks ago. i.e. #1822 and #1955 can both be potential fix to the problem described in this issue

Is this issue encountered for all models? To be honest the `Cache.add()` is a bit vague; I've encountered this when the URL is wrong, but it does not seem to...