Junru Shao
Junru Shao
Currently we are working on the documentation page: https://mlc.ai/mlc-llm/docs/. Still lots of work to do. Closing this issue as it's a bit generic. Please do bring up specific doc requests...
CC @Hzfengsy
This is weird because I did test it with my iPhone 14 pro, and it seems to work. The only known reason so far about crashes are due to OOM,...
Looks like upgrading iOS to the latest should work
@logan-markewich does it work now with your AMD GPU? If so, would you mind sharing your tok/sec info [here](https://github.com/mlc-ai/mlc-llm/issues/15)?
I just bought an AMD GPU online. It's arriving later this week. Will play with it and see how it works with the Vulkan driver.
@nRuaif 4GB memory wouldn't be enough. A 6GB one should work
@y-lee That's correct. The model we are using so far requires 6GB RAM to run smoothly
@zifken looks like `VK_ERROR_OUT_OF_DEVICE_MEMORY` indicates that it doesn't have enough memory. I looked it up and it seems that GT 920M only has 2GB RAM, but the default model is...
@zifken there are some reports saying 4GB might work, but 6GB is recommended atm