sens

edit your local file "C:\Users\user\AppData\Local\Programs\Python\Python310\lib\site-packages\objection\agent.js", find this lines: ``` return { activity: currentActivity.$className, fragment: fragment.$className, }; ``` modifiy to: ``` return { activity: currentActivity.$className, fragment: fragment?.$className ?? null, }; ```

@mengshyu 3-bit quantization is available, but it is particularly slow, with only about 0.7 tok/s prefill and 1.1 tok/s decode.

It seems that this level of hardware configuration is not yet able to drive models of 8B and above. Currently, I have found that only models around 1.5B can run...

> Hi, thanks for your interest in SDAI app. > > The Local Diffusion functionality is still in the experimental stage, so the structure of the package or the interface...

Modify the `overrides` parameter when [Launch the Server](https://llm.mlc.ai/docs/deploy/rest.html#id4)

> > Modify the `overrides` parameter when [Launch the Server](https://llm.mlc.ai/docs/deploy/rest.html#id4) > > I can modify "gpu_memory_utilization" in “mlc_llm serve” mode. How to set it when using "mlc_llm chat"? `mlc_lm chat`...

> > 大佬你好，建议增加一个可以快速接入到Android的WebView中的库，方便使用WebView打开网址也可以使用猫爪进行资源嗅探下载。 > > 不太可能，我不知道是否有证据表明这一点。 > > 建议搭配kiwi浏览器。 仅实现嗅探功能也不行吗？只需要识别出资源的下载地址，具体的下载可以交给外部去实现。毕竟很多时候"能下载"的需求大于"快速下载"的。