Naming Conventions【免费下载链接】geGEGraph Engine是面向昇腾的图编译器和执行器提供了计算图优化、多流并行、内存复用和模型下沉等技术手段加速模型执行效率减少模型内存占用。 GE 提供对 PyTorch、TensorFlow 前端的友好接入能力并同时支持 onnx、pb 等主流模型格式的解析与编译。项目地址: https://gitcode.com/cann/ge1. File NamingSplit files and name them at the feature granularity levelOK: test_ctrl_flow_infershape.cc This indicates that this group of test cases primarily tests control flow infershape-related features. **NOK: test_graph_compiler.cc** Its unclear what feature is being tested.2. Test Case NamingTest case names need to clearly indicate the testing pointOK: TEST_F(ConstantFoldingTest, test_flattenv2_folding_success) This indicates that this test case tests the normal scenario of flattenv2s constant folding functionality **NOK: TEST_F(FftsPlusTest, ffts_plus_graph)** Its unclear what functionality is being tested.Verification Points1. Test cases must have verification points; empty flow test cases are not allowed// build_graph through sessionret session.BuildGraph(2, inputs);EXPECT_EQ(ret, SUCCESS);Only verified the interface return success, with no other verification points. This basically has no guarding function.2. Test cases need to be end-to-end entry test cases; copying UT test cases is not allowed【免费下载链接】geGEGraph Engine是面向昇腾的图编译器和执行器提供了计算图优化、多流并行、内存复用和模型下沉等技术手段加速模型执行效率减少模型内存占用。 GE 提供对 PyTorch、TensorFlow 前端的友好接入能力并同时支持 onnx、pb 等主流模型格式的解析与编译。项目地址: https://gitcode.com/cann/ge创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考