Fix GLM4 incoherence with fp16 accumulators#13639

Merged

0cc4m merged 1 commit intomasterfrom

0cc4m/fix-vulkan-glm4

May 20, 2025

Commits on May 19, 2025

Set GLM4 blk.*.attn_output.weight, kqv_out-* matmul to GGML_PREC_F32 to fix infinity values in output
0cc4m
committed