PrivateGPT on AVX only system

[anantshri]

I have been experimenting with older CPU's. Powerful CPU's (10 core dual cpu unit so total of 40 core available) but they are old say 5-6 years old and hence AVX2 support is non existent so many of the privategpt/gpt4all stuff doesn't work at all.

I have documented what worked and how:

TL:DR; llama models work, remember to requantise them if you have old model and if you do that remember to upgrade your dependencies.

full details https://blog.anantshri.info/privategpt-and-cpus-with-no-avx2/