From 7250c1ec0f8e02c53af27c2ae9c70a518fb41cc4 Mon Sep 17 00:00:00 2001 From: Gorokhov Dmitriy Date: Mon, 14 Oct 2024 09:58:38 +0400 Subject: [PATCH] [CPU] Enabled FP16 Compressed FC on models with PagedAttention (#26279) --- src/plugins/intel_cpu/src/plugin.cpp | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/src/plugins/intel_cpu/src/plugin.cpp b/src/plugins/intel_cpu/src/plugin.cpp index 1082157e86d53f..fa1810ff6044f9 100644 --- a/src/plugins/intel_cpu/src/plugin.cpp +++ b/src/plugins/intel_cpu/src/plugin.cpp @@ -19,6 +19,7 @@ #include "utils/precision_support.h" #include "utils/serialize.hpp" #include "weights_cache.hpp" +#include "openvino/op/paged_attention.hpp" #if defined(__linux__) # include @@ -197,9 +198,9 @@ static Config::ModelType getModelType(const std::shared_ptr& model) if (op::util::has_op_with_type(model) || op::util::has_op_with_type(model)) return Config::ModelType::CNN; - - if (op::util::has_op_with_type(model) && - model->get_variables().size() > 0) + + if ((op::util::has_op_with_type(model) && model->get_variables().size() > 0) || + op::util::has_op_with_type(model)) return Config::ModelType::LLM; return Config::ModelType::Unknown;