caffe训练出现问题:
I0417 14:05:52.033957 2040 net.cpp:226] Scale1 needs backward computation.
I0417 14:05:52.033962 2040 net.cpp:226] BatchNorm1 needs backward computation.I0417 14:05:52.033967 2040 net.cpp:226] Convolution1 needs backward computation.
I0417 14:05:52.033975 2040 net.cpp:228] Data2_Data1_1_split does not need backward computation.
I0417 14:05:52.033982 2040 net.cpp:228] Data1 does not need backward computation.
I0417 14:05:52.033987 2040 net.cpp:270] This network produces output Accuracy1
I0417 14:05:52.033994 2040 net.cpp:270] This network produces output SoftmaxWithLoss1
I0417 14:05:52.034227 2040 net.cpp:283] Network initialization done.
I0417 14:05:52.035884 2040 solver.cpp:60] Solver scaffolding done.
I0417 14:05:52.062779 2040 caffe.cpp:251] Starting Optimization
I0417 14:05:52.062799 2040 solver.cpp:279] Solving resnet_cifar10
I0417 14:05:52.062822 2040 solver.cpp:280] Learning Rate Policy: multistep
I0417 14:05:52.093228 2040 solver.cpp:337] Iteration 0, Testing net (#0)
F0417 14:05:52.344153 2040 syncedmem.cpp:56] Check failed: error == cudaSuccess (2 vs. 0) out of memory
*** Check failure stack trace: ***
@ 0x7f83ef095daa (unknown)
@ 0x7f83ef095ce4 (unknown)
@ 0x7f83ef0956e6 (unknown)
@ 0x7f83ef098687 (unknown)
@ 0x7f83ef7be581 caffe::SyncedMemory::to_gpu()
@ 0x7f83ef7bd8e9 caffe::SyncedMemory::mutable_gpu_data()
@ 0x7f83ef6a7a92 caffe::Blob<>::mutable_gpu_data()
@ 0x7f83ef845484 caffe::ScaleLayer<>::Forward_gpu()
@ 0x7f83ef7c51c5 caffe::Net<>::ForwardFromTo()
@ 0x7f83ef7c5537 caffe::Net<>::Forward()
@ 0x7f83ef7eff8c caffe::Solver<>::Test()
@ 0x7f83ef7f08be caffe::Solver<>::TestAll()
@ 0x7f83ef7f09b0 caffe::Solver<>::Step()
@ 0x7f83ef7f1379 caffe::Solver<>::Solve()
@ 0x40873b train()
@ 0x405b3c main
@ 0x7f83ee0a1ec5 (unknown)
@ 0x4063ab (unknown)
@ (nil) (unknown)
解决:
batchsize改成2