使用cuda进行测试，Cannot add vertex: a vertex with name "outputs" already exists
xiaoyangmoa opened this issue · 2 comments
xiaoyangmoa commented
cuda各个版本为,cpu版本运行正常，但是速度太慢了:
<nd4j.version>1.0.0-beta3</nd4j.version>
<dl4j.version>1.0.0-beta3</dl4j.version>
<datavec.version>1.0.0-beta3</datavec.version>
<arbiter.version>1.0.0-beta3</arbiter.version>
<rl4j.version>1.0.0-beta3</rl4j.version>
运行报错为:
java.lang.IllegalStateException: Cannot add vertex: a vertex with name "outputs" already exists
at org.nd4j.base.Preconditions.throwStateEx(Preconditions.java:641)
at org.nd4j.base.Preconditions.checkState(Preconditions.java:304)
at org.deeplearning4j.nn.conf.ComputationGraphConfiguration$GraphBuilder.addVertex(ComputationGraphConfiguration.java:924)
at org.deeplearning4j.nn.conf.ComputationGraphConfiguration$GraphBuilder.addLayer(ComputationGraphConfiguration.java:788)
at org.deeplearning4j.nn.transferlearning.TransferLearning$GraphBuilder.addLayer(TransferLearning.java:817)
at App.train(App.java:143)
at App.main(App.java:77)
xiaoyangmoa commented
RedBloodCellDetection 对应这个文件里面的 outputs 的层出错，修改了会出现其它错误
sjsdfg commented
问题无法复现
请自己检查配置
o.RedBloodCellDetection - Load data...
o.RedBloodCellDetection - Build model...
o.d.z.ZooModel - Using cached model at C:\Users\73677\.deeplearning4j\models\tinyyolo\tiny-yolo-voc_dl4j_inference.v2.zip
o.d.z.ZooModel - Verifying download...
o.d.z.ZooModel - Checksum local is 1256226465, expecting 1256226465
o.n.l.f.Nd4jBackend - Loaded [JCublasBackend] backend
o.n.n.NativeOpsHolder - Number of threads used for NativeOps: 32
o.n.n.Nd4jBlas - Number of threads used for BLAS: 0
o.n.l.a.o.e.DefaultOpExecutioner - Backend used: [CUDA]; OS: [Windows 10]
o.n.l.a.o.e.DefaultOpExecutioner - Cores: [8]; Memory: [3.5GB];
o.n.l.a.o.e.DefaultOpExecutioner - Blas vendor: [CUBLAS]
o.n.l.j.o.e.CudaExecutioner - Device Name: [GeForce GTX 1050 Ti]; CC: [6.1]; Total/free memory: [4294967296]
o.d.n.g.ComputationGraph - Starting ComputationGraph with WorkspaceModes set to [training: NONE; inference: SEPARATE], cacheMode set to [NONE]
==========================================================================================================================================================================================================================================================
VertexName (VertexType) nIn,nOut TotalParams ParamsShape Vertex Inputs InputShape OutputShape
==========================================================================================================================================================================================================================================================
input_1 (InputVertex) -,- - - - - -
conv2d_1 (ConvolutionLayer) 3,16 432 W:{16,3,3,3} [input_1] InputTypeConvolutional(h=416,w=416,c=3) InputTypeConvolutional(h=416,w=416,c=16)
batch_normalization_1 (BatchNormalization)16,16 64 gamma:{1,16}, beta:{1,16}, mean:{1,16}, var:{1,16}[conv2d_1] InputTypeConvolutional(h=416,w=416,c=16) InputTypeConvolutional(h=416,w=416,c=16)
leaky_re_lu_1 (ActivationLayer) -,- 0 - [batch_normalization_1] InputTypeConvolutional(h=416,w=416,c=16) InputTypeConvolutional(h=416,w=416,c=16)
max_pooling2d_1 (SubsamplingLayer) -,- 0 - [leaky_re_lu_1] InputTypeConvolutional(h=416,w=416,c=16) InputTypeConvolutional(h=208,w=208,c=16)
conv2d_2 (ConvolutionLayer) 16,32 4608 W:{32,16,3,3} [max_pooling2d_1] InputTypeConvolutional(h=208,w=208,c=16) InputTypeConvolutional(h=208,w=208,c=32)
batch_normalization_2 (BatchNormalization)32,32 128 gamma:{1,32}, beta:{1,32}, mean:{1,32}, var:{1,32}[conv2d_2] InputTypeConvolutional(h=208,w=208,c=32) InputTypeConvolutional(h=208,w=208,c=32)
leaky_re_lu_2 (ActivationLayer) -,- 0 - [batch_normalization_2] InputTypeConvolutional(h=208,w=208,c=32) InputTypeConvolutional(h=208,w=208,c=32)
max_pooling2d_2 (SubsamplingLayer) -,- 0 - [leaky_re_lu_2] InputTypeConvolutional(h=208,w=208,c=32) InputTypeConvolutional(h=104,w=104,c=32)
conv2d_3 (ConvolutionLayer) 32,64 18432 W:{64,32,3,3} [max_pooling2d_2] InputTypeConvolutional(h=104,w=104,c=32) InputTypeConvolutional(h=104,w=104,c=64)
batch_normalization_3 (BatchNormalization)64,64 256 gamma:{1,64}, beta:{1,64}, mean:{1,64}, var:{1,64}[conv2d_3] InputTypeConvolutional(h=104,w=104,c=64) InputTypeConvolutional(h=104,w=104,c=64)
leaky_re_lu_3 (ActivationLayer) -,- 0 - [batch_normalization_3] InputTypeConvolutional(h=104,w=104,c=64) InputTypeConvolutional(h=104,w=104,c=64)
max_pooling2d_3 (SubsamplingLayer) -,- 0 - [leaky_re_lu_3] InputTypeConvolutional(h=104,w=104,c=64) InputTypeConvolutional(h=52,w=52,c=64)
conv2d_4 (ConvolutionLayer) 64,128 73728 W:{128,64,3,3} [max_pooling2d_3] InputTypeConvolutional(h=52,w=52,c=64) InputTypeConvolutional(h=52,w=52,c=128)
batch_normalization_4 (BatchNormalization)128,128 512 gamma:{1,128}, beta:{1,128}, mean:{1,128}, var:{1,128}[conv2d_4] InputTypeConvolutional(h=52,w=52,c=128) InputTypeConvolutional(h=52,w=52,c=128)
leaky_re_lu_4 (ActivationLayer) -,- 0 - [batch_normalization_4] InputTypeConvolutional(h=52,w=52,c=128) InputTypeConvolutional(h=52,w=52,c=128)
max_pooling2d_4 (SubsamplingLayer) -,- 0 - [leaky_re_lu_4] InputTypeConvolutional(h=52,w=52,c=128) InputTypeConvolutional(h=26,w=26,c=128)
conv2d_5 (ConvolutionLayer) 128,256 294912 W:{256,128,3,3} [max_pooling2d_4] InputTypeConvolutional(h=26,w=26,c=128) InputTypeConvolutional(h=26,w=26,c=256)
batch_normalization_5 (BatchNormalization)256,256 1024 gamma:{1,256}, beta:{1,256}, mean:{1,256}, var:{1,256}[conv2d_5] InputTypeConvolutional(h=26,w=26,c=256) InputTypeConvolutional(h=26,w=26,c=256)
leaky_re_lu_5 (ActivationLayer) -,- 0 - [batch_normalization_5] InputTypeConvolutional(h=26,w=26,c=256) InputTypeConvolutional(h=26,w=26,c=256)
max_pooling2d_5 (SubsamplingLayer) -,- 0 - [leaky_re_lu_5] InputTypeConvolutional(h=26,w=26,c=256) InputTypeConvolutional(h=13,w=13,c=256)
conv2d_6 (ConvolutionLayer) 256,512 1179648 W:{512,256,3,3} [max_pooling2d_5] InputTypeConvolutional(h=13,w=13,c=256) InputTypeConvolutional(h=13,w=13,c=512)
batch_normalization_6 (BatchNormalization)512,512 2048 gamma:{1,512}, beta:{1,512}, mean:{1,512}, var:{1,512}[conv2d_6] InputTypeConvolutional(h=13,w=13,c=512) InputTypeConvolutional(h=13,w=13,c=512)
leaky_re_lu_6 (ActivationLayer) -,- 0 - [batch_normalization_6] InputTypeConvolutional(h=13,w=13,c=512) InputTypeConvolutional(h=13,w=13,c=512)
max_pooling2d_6 (SubsamplingLayer) -,- 0 - [leaky_re_lu_6] InputTypeConvolutional(h=13,w=13,c=512) InputTypeConvolutional(h=13,w=13,c=512)
conv2d_7 (ConvolutionLayer) 512,1024 4718592 W:{1024,512,3,3} [max_pooling2d_6] InputTypeConvolutional(h=13,w=13,c=512) InputTypeConvolutional(h=13,w=13,c=1024)
batch_normalization_7 (BatchNormalization)1024,1024 4096 gamma:{1,1024}, beta:{1,1024}, mean:{1,1024}, var:{1,1024}[conv2d_7] InputTypeConvolutional(h=13,w=13,c=1024) InputTypeConvolutional(h=13,w=13,c=1024)
leaky_re_lu_7 (ActivationLayer) -,- 0 - [batch_normalization_7] InputTypeConvolutional(h=13,w=13,c=1024) InputTypeConvolutional(h=13,w=13,c=1024)
conv2d_8 (ConvolutionLayer) 1024,1024 9437184 W:{1024,1024,3,3} [leaky_re_lu_7] InputTypeConvolutional(h=13,w=13,c=1024) InputTypeConvolutional(h=13,w=13,c=1024)
batch_normalization_8 (BatchNormalization)1024,1024 4096 gamma:{1,1024}, beta:{1,1024}, mean:{1,1024}, var:{1,1024}[conv2d_8] InputTypeConvolutional(h=13,w=13,c=1024) InputTypeConvolutional(h=13,w=13,c=1024)
leaky_re_lu_8 (ActivationLayer) -,- 0 - [batch_normalization_8] InputTypeConvolutional(h=13,w=13,c=1024) InputTypeConvolutional(h=13,w=13,c=1024)
convolution2d_9 (ConvolutionLayer) 1024,30 30720 W:{30,1024,1,1} [leaky_re_lu_8] InputTypeConvolutional(h=13,w=13,c=1024) InputTypeConvolutional(h=13,w=13,c=30)
outputs (Yolo2OutputLayer) -,- 0 - [convolution2d_9] InputTypeConvolutional(h=13,w=13,c=30) InputTypeConvolutional(h=13,w=13,c=30)
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Total Parameters: 15770480
Trainable Parameters: 15770480
Frozen Parameters: 0
==========================================================================================================================================================================================================================================================
o.RedBloodCellDetection - Train model...
o.d.o.l.ScoreIterationListener - Score at iteration 0 is 144.8436877360403
o.d.o.l.ScoreIterationListener - Score at iteration 1 is 136.6377426532769
o.d.o.l.ScoreIterationListener - Score at iteration 2 is 146.40208036328607
o.d.o.l.ScoreIterationListener - Score at iteration 3 is 140.99282526824803
o.d.o.l.ScoreIterationListener - Score at iteration 4 is 144.56626546857055
o.d.o.l.ScoreIterationListener - Score at iteration 5 is 147.27808616722498