我在训练一个网络，并将学习率从0.1调整到了0.00001。输出始终保持不变。训练过程中没有使用均值。导致这种奇怪损失的原因可能有哪些？

I1107 15:07:28.381621 12333 solver.cpp:404]     Test net output #0: loss = 3.37134e+11 (* 1 = 3.37134e+11 loss)I1107 15:07:28.549142 12333 solver.cpp:228] Iteration 0, loss = 1.28092e+11I1107 15:07:28.549201 12333 solver.cpp:244]     Train net output #0: loss = 1.28092e+11 (* 1 = 1.28092e+11 loss)I1107 15:07:28.549211 12333 sgd_solver.cpp:106] Iteration 0, lr = 1e-07I1107 15:07:59.490077 12333 solver.cpp:228] Iteration 50, loss = -nanI1107 15:07:59.490170 12333 solver.cpp:244]     Train net output #0: loss = 0 (* 1 = 0 loss)I1107 15:07:59.490176 12333 sgd_solver.cpp:106] Iteration 50, lr = 1e-07I1107 15:08:29.177093 12333 solver.cpp:228] Iteration 100, loss = -nanI1107 15:08:29.177119 12333 solver.cpp:244]     Train net output #0: loss = 0 (* 1 = 0 loss)I1107 15:08:29.177125 12333 sgd_solver.cpp:106] Iteration 100, lr = 1e-07I1107 15:08:59.758381 12333 solver.cpp:228] Iteration 150, loss = -nanI1107 15:08:59.758513 12333 solver.cpp:244]     Train net output #0: loss = 0 (* 1 = 0 loss)I1107 15:08:59.758545 12333 sgd_solver.cpp:106] Iteration 150, lr = 1e-07I1107 15:09:30.210208 12333 solver.cpp:228] Iteration 200, loss = -nanI1107 15:09:30.210304 12333 solver.cpp:244]     Train net output #0: loss = 0 (* 1 = 0 loss)I1107 15:09:30.210310 12333 sgd_solver.cpp:106] Iteration 200, lr = 1e-07

回答：

你的损失并非0，甚至相差甚远。你开始时的损失是3.3e+11（约10^11），看起来很快就爆炸了，变成了nan。你需要大幅度缩小你的损失值。如果你使用的是"EuclideanLoss"，你可能需要根据深度图的大小来平均损失，将预测值缩放到[-1,1]范围，或者使用其他任何可以防止损失爆炸的缩放方法。