YarrMatey(she/her)

I'm vegan

Pirate jamming to stereo

wiki-user: YarrMatey

  • 0 Posts
  • 6 Comments
Joined 3 years ago
cake
Cake day: August 16th, 2023

help-circle
  • Thing is, my specs are really good. Again, my laptop is a beast and it ran like 100% CPU, 100% GPU, and 100% of my 64gb of DDR5 Ram! Seeing it and how hot it made my computer

    Local llm usage depends on your VRAM, your GPU. It sounds like you tried to run a model your laptop can’t handle. If you also need your RAM, then it is going to be super slow no matter what. The number each model has is important to determine if your GPU can handle it, also the quantization. If you run something significantly smaller like 2B* rather than say gemma 31B, then a laptop might be able to handle it. Your laptop is not a beast (for LLM usage) unless it has more than 24 GB of VRAM on it so don’t assume you can run any model on it. Look for smaller numbers with quantization. Running a local llm for me on my desktop is similar to running a video game, it doesn’t slow to a crawl or cause 100% on CPU or RAM because only the VRAM is touched.

    *or 4B even, just depends on your GPU