你好，训练自己数据的时候准确率和loss都为0

[LimkeeTan]

你好，我在用自己数据训练的时候，出现准确率和loss都为0的情况，我的训练数据是1050张，测试数据是50张，如下，我一开始以为是batch太大了，所以我改成64，但还是这样，请问是为什么
I0505 20:36:57.910152 15093 solver.cpp:331] Iteration 0, Testing net (#0)
I0505 20:36:58.462410 15093 solver.cpp:398] Test net output #0: accuracy = 0
I0505 20:36:58.462445 15093 solver.cpp:398] Test net output #1: ctc_loss = 0.0889562 (* 1 = 0.0889562 loss)
I0505 20:36:58.503353 15093 solver.cpp:219] Iteration 0 (-2.35495e-09 iter/s, 0.593967s/10 iters), loss = 0
I0505 20:36:58.503383 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:36:58.503389 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:36:58.503413 15093 sgd_solver.cpp:105] Iteration 0, lr = 0.001
I0505 20:36:58.832192 15093 solver.cpp:219] Iteration 10 (30.4131 iter/s, 0.328806s/10 iters), loss = 0
I0505 20:36:58.832212 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:36:58.832217 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:36:58.832235 15093 sgd_solver.cpp:105] Iteration 10, lr = 0.001
I0505 20:36:59.160102 15093 solver.cpp:219] Iteration 20 (30.4983 iter/s, 0.327887s/10 iters), loss = 0
I0505 20:36:59.160135 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:36:59.160140 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:36:59.160143 15093 sgd_solver.cpp:105] Iteration 20, lr = 0.001
I0505 20:36:59.487865 15093 solver.cpp:219] Iteration 30 (30.5132 iter/s, 0.327727s/10 iters), loss = 0
I0505 20:36:59.487886 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:36:59.487892 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:36:59.487895 15093 sgd_solver.cpp:105] Iteration 30, lr = 0.001
I0505 20:36:59.814749 15093 solver.cpp:219] Iteration 40 (30.5943 iter/s, 0.326858s/10 iters), loss = 0
I0505 20:36:59.814783 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:36:59.814790 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:36:59.814839 15093 sgd_solver.cpp:105] Iteration 40, lr = 0.001
I0505 20:37:00.145546 15093 solver.cpp:219] Iteration 50 (30.2332 iter/s, 0.330763s/10 iters), loss = 0
I0505 20:37:00.145583 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:37:00.145589 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:37:00.145593 15093 sgd_solver.cpp:105] Iteration 50, lr = 0.001
I0505 20:37:00.474901 15093 solver.cpp:219] Iteration 60 (30.3663 iter/s, 0.329312s/10 iters), loss = 0
I0505 20:37:00.474920 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:37:00.474926 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:37:00.474931 15093 sgd_solver.cpp:105] Iteration 60, lr = 0.001
I0505 20:37:00.800871 15093 solver.cpp:219] Iteration 70 (30.6796 iter/s, 0.32595s/10 iters), loss = 0
I0505 20:37:00.800900 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:37:00.800906 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:37:00.800909 15093 sgd_solver.cpp:105] Iteration 70, lr = 0.001
I0505 20:37:01.128326 15093 solver.cpp:219] Iteration 80 (30.5412 iter/s, 0.327426s/10 iters), loss = 0
I0505 20:37:01.128342 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:37:01.128348 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:37:01.128367 15093 sgd_solver.cpp:105] Iteration 80, lr = 0.001
I0505 20:37:01.455250 15093 solver.cpp:219] Iteration 90 (30.5905 iter/s, 0.326899s/10 iters), loss = 0
I0505 20:37:01.455288 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:37:01.455296 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:37:01.455301 15093 sgd_solver.cpp:105] Iteration 90, lr = 0.001
I0505 20:37:01.783249 15093 solver.cpp:219] Iteration 100 (30.4918 iter/s, 0.327956s/10 iters), loss = 0
I0505 20:37:01.783283 15093 solver.cpp:238] Train net output #0: accuracy = 0
I0505 20:37:01.783289 15093 solver.cpp:238] Train net output #1: ctc_loss = 0 (* 1 = 0 loss)
I0505 20:37:01.783293 15093 sgd_solver.cpp:105] Iteration 100, lr = 0.001

[YangPengYu310]

请问在你的项目中有一个混合网格的生成，你那里是否有任何混合网格的vtk文件呢