[m1sta_]

This looks really really interesting. Any idea whether it would run on a laptop with an Intel Core i7?

[Const-me]

The performance on integrated GPUs is not stellar, but it should work.

On my AMD Ryzen 5 5600u with dual-channel DDR4, I’m getting 2 tokens/second. My friend with Intel Core i3 and single-channel memory was getting 1 token/second.

[MacsHeadroom]

Yes, I run it on CPU using LLMStudio. It's very fast.

[ru552]

VRAM is king. If you have the VRAM to hold the model parameters, you can run it.