Aaron Miller

ah yeah it seems like it just silently fails to run the model when hitting a memory limit when running under `lldb` you actually get warnings about it ``` 2023-06-15...

It appears to just be a memory size thing, smaller models take a bigger context size to start breaking, k-quant or not

> Should I make a pull request for checking if the command buffer failed and throwing an error when it has insufficient memory? I think its worth it - crashing...

yes - this is fixed now that this crashes instead of giving bad output