Fix mixed-bit quantized load for sanitize()-derived MLA projections - #1482

Merged

michalk8 merged 6 commits into

ml-explore:mainfrom

rajanshxrma:fix-mixed-bit-mla-derived-quantization

Sep 18, 2026

Commits on Jul 6, 2026

fix: infer bits/group_size for quantized paths missing from the per-tensor map
rajanshxrma
committed
1c9597a

Commits on Sep 18, 2026

Merge branch 'main' into pr-1482
michalk8
committed
e381bc4
Add `infer_quant_config`
michalk8
committed
d49377b
Add tests
michalk8
committed
7a6ed5c
Update test
michalk8
committed
eb49c3c
Merge branch 'main' into fix-mixed-bit-mla-derived-quantization
michalk8
authored
ef325cb