diff --git a/onnxruntime/core/providers/openvino/ov_interface.cc b/onnxruntime/core/providers/openvino/ov_interface.cc index 23be3447b8799..eb0f201643060 100644 --- a/onnxruntime/core/providers/openvino/ov_interface.cc +++ b/onnxruntime/core/providers/openvino/ov_interface.cc @@ -552,7 +552,6 @@ void StatefulOVInferRequest::RewindKVCache(size_t index) { } OVTensorPtr StatefulOVInferRequest::GetTensor(const std::string& input_name) { - auto tobj = OVInferRequest::GetTensor(input_name); if (_npu_logits_slice_required) { @@ -583,7 +582,7 @@ OVTensorPtr StatefulOVInferRequest::GetTensor(const std::string& input_name) { return blob; } }, - "Could not create sliced logits tensor"); + "Could not create sliced logits tensor"); } } } diff --git a/onnxruntime/core/providers/openvino/ov_stateful_patch_utils.cc b/onnxruntime/core/providers/openvino/ov_stateful_patch_utils.cc index c4ec47534d009..fd2b5797a1f40 100644 --- a/onnxruntime/core/providers/openvino/ov_stateful_patch_utils.cc +++ b/onnxruntime/core/providers/openvino/ov_stateful_patch_utils.cc @@ -178,7 +178,6 @@ std::pair, std::unordered_set> ExtractKVPa // not_kv_inputs = ["input_ids", "attention_mask"] std::pair, std::vector> ExtractInputKVTensors( const std::shared_ptr& model, const std::unordered_set& kv_patterns) { - std::vector key_value_input_names; std::vector not_kv_inputs; @@ -258,11 +257,11 @@ void PatchStatefulDecoder(std::shared_ptr model) { } if (key_value_input_names.size() != key_value_output_names.size()) { - ORT_THROW("Found different sizes between key_value_input_names (", - key_value_input_names.size(), - ") and key_value_output_names (", - key_value_output_names.size(), - "). They couldn't be paired."); + ORT_THROW("Found different sizes between key_value_input_names (", + key_value_input_names.size(), + ") and key_value_output_names (", + key_value_output_names.size(), + "). They couldn't be paired."); } // By default, batch is the 0 - th but chatglm uses 1 - st dimension as batch