sandyhsia

This repo is not even training the SE layer, since the author didn't put the ops into nn.Sequential(). That means, your 'squeeze and excitation' will only depend on what's randomly...

TEST_CASE("test from issues") { test test1; std::string str1 = R"({"username1": "test", "password":test, "id": 10.1, "error": false})"; iguana::from_json(test1, str1.c_str(), str1.length()); std::cout