Skip to content

Commit 1ce0653

Browse files
authored
[INTEL_HPU] change fused_quant scale output format to bf16 (#1850)
1 parent e6d0348 commit 1ce0653

File tree

1 file changed

+1
-1
lines changed

1 file changed

+1
-1
lines changed

backends/intel_hpu/custom_ops/llama_infer/fused_quant.cc

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -174,7 +174,7 @@ void fused_quant(const Context& dev_ctx,
174174
return;
175175
}
176176

177-
dev_ctx.template Alloc<float>(scale);
177+
dev_ctx.template Alloc<phi::dtype::bfloat16>(scale);
178178
if (scale->numel() == 0) {
179179
return;
180180
}

0 commit comments

Comments
 (0)