WebGPU in Chrome Canary now supports float16 for faster local LLM inference

WebGPU in Chrome Canary now supports float16 for faster local LLM inference(twitter.com)

18 points by NiekvdMaas 2 years ago | 5 comments

no_time 2 years ago |

I hope they gate webGPU behind a per domain allow/deny notification like they do with location and camera access.

This whole thing is just begging to be exploited by cryptominers.

charcircuit 2 years ago | |

It may be more profitable to use the compute for LLMs or other GPU dependent AI than crypto mining. The downside is that there is no way to verify the client isn't returning random or bad data.

lostmsu 2 years ago | |

Cryptomining is mostly gone.

NiekvdMaas 2 years ago |

To test, run Chrome Canary with:

  --enable-dawn-features=allow_unsafe_apis,disable_robustness

Then, open: https://webllm.mlc.ai/#chat-demo

hardware2win 2 years ago | |

>disable_robustness

What a name, especially since it improves perf