os360

Bug: Crash in FlashQKV<128, 8, 128>::normalize_and_store_1row failing with GLM 4.5, AVX2, -amb/-ub/-b

5

### What happened? I was comparing the output of DeepSeek vs GLM-4.5 when I isolated a case where llama-server repeatedly fails when these parameters are passed: --attention-max-batch 2048 --batch-size 16384...