doldolggi/EvoCUA-8B-20260105-GGUF

Fix: test targets are 46-232px (not 100-232) - one 46px toggle exists; add its per-tier drift

Fix: test targets are 46-232px (not 100-232) - one 46px toggle exists; add its per-tier drift

Card: link the 32B sibling repo (weights now published)

Card: explain why 32B measurements are included but 32B weights are not

Add GGUF quants (Q8_0..Q3_K_M, UD-Q5/Q4/Q3) + F16 mmproj + imatrix

Add model card, coordinate-regression harness, and raw results

initial commit