Upload qwen3.6-27b.mq3 with huggingface_hub

Upload qwen36-27b-dflash-mq3.hfq with huggingface_hub

Add v3 TriAttention sidecar (blended 1M corpus, R̄=0.610)

model card: document files, draft pairing, 2026-04-27 refresh

refresh draft: z-lab/Qwen3.6-27B-DFlash@0919688 (2026-04-27); +4-11% tok/s vs 1dbb59a

Upload qwen3.6-27b.mq4 with huggingface_hub

Upload qwen36-27b-dflash-mq4.hfq with huggingface_hub

initial commit