feat(MODEL-MM-QWEN4-EXP): W6a — IQ4_NL, Q5_0 and a dequantizing gather, so the only artifact that fits can be opened - #2019

Open

localai-bot wants to merge 11 commits into

mainfrom

row/MODEL-MM-QWEN4-EXP-W6A

Commits on Aug 26, 2026

feat(MODEL-MM-QWEN4-EXP): W6a — IQ4_NL, a dequantizing gather, and the qwen4exp config builder (#1989)
mudler
committed
beedfdf
fix(MODEL-MM-QWEN4-EXP): W6a — two GGUF loaders could not read the table the shared policy started keeping
mudler
committed
6bddb91
fix(MODEL-MM-QWEN4-EXP): W6a — gate the two vec_dots, the two refusals, and the bridge the forward actually uses
mudler
committed
aa02dbc
record(MODEL-MM-QWEN4-EXP): W6a is the first wave to land, so it owes the row transition and three corrections
mudler
committed
7109ef1
Merge origin/main into row/MODEL-MM-QWEN4-EXP-W6A
mudler
committed
a1d414f
record(MODEL-MM-QWEN4-EXP): W1 and W6a both moved this row to ACTIVE, and only the counters survive that by luck
mudler
committed
19d1b06
record(MODEL-MM-QWEN4-EXP): the ACTIVE transition owed a ratchet-test update, two anchors and a row-ID mention
mudler
committed
08d1aab
merge: origin/main into row/MODEL-MM-QWEN4-EXP-W6A
mudler
committed
4f1962f
fix(MODEL-MM-QWEN4-EXP): the spec's settled-decisions list said the OPPOSITE of the row's settled QSA mapping
mudler
committed
3d31a57
merge: origin/main into row/MODEL-MM-QWEN4-EXP-W6A (second pass)
mudler
committed
7e8344a

Commits on Aug 27, 2026

merge: origin/main into row/MODEL-MM-QWEN4-EXP-W6A (third pass)
mudler
committed
2481a3a