--iq2auto

Upload 4 files

IQ2_XS

optimizations for normal llama

Fold test

Fold baked into standard Q2K codebook

DC cancellation buff

Prototype build

I prefer Q2 honestly.. but leaving IQ1 staging in for posterity

IQ1

Lower RMSE, improve error shaping

Exponential speed-up

--analog-imatrix: Reduce error in output by 15%

Replace Shor's with accelerated sieve

Update hexstate_quantize.c

Update hexstate_quantize.c

Q8_0 tied embeddings

Update hexstate_quantize.c

Revert to Alpha 0.1

ALPHA

ALPHA

ALPHA

Upload 5 files

Qwen changes

Heavily experimental

Experimental imatrix

Experimental support for other LLMs

It's only calibrated for Gemma, atm.