From 85c32fdd1034288402813366413dfcc415cd6bdb Mon Sep 17 00:00:00 2001 From: Azure-Tang Date: Sat, 15 Mar 2025 22:27:02 -0400 Subject: [PATCH] Fix rocm example yaml --- .../optimize/optimize_rules/rocm/DeepSeek-V3-Chat.yaml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ktransformers/optimize/optimize_rules/rocm/DeepSeek-V3-Chat.yaml b/ktransformers/optimize/optimize_rules/rocm/DeepSeek-V3-Chat.yaml index 628a952..ffbff2b 100644 --- a/ktransformers/optimize/optimize_rules/rocm/DeepSeek-V3-Chat.yaml +++ b/ktransformers/optimize/optimize_rules/rocm/DeepSeek-V3-Chat.yaml @@ -12,7 +12,7 @@ replace: class: ktransformers.operators.linear.KTransformersLinear # optimized Kernel on quantized data types kwargs: - generate_device: "cuda" + generate_device: "cpu" prefill_device: "cuda" generate_op: "KLinearCPUInfer" prefill_op: "KLinearTorch" @@ -23,7 +23,7 @@ replace: class: ktransformers.operators.linear.KTransformersLinear # optimized Kernel on quantized data types kwargs: - generate_device: "cpu" + generate_device: "cuda" prefill_device: "cuda" generate_op: "KLinearQ8" prefill_op: "KLinearTorch"