View Code? Open in Web Editor
NEW
OpenCL for Nets - A Deep Learning Framework based on OpenCL, written by C++. Supports popular MLP, RNN(LSTM), CNN(ResNet). Friendly debugger. Transparent data. No library dependencies. 基于OpenCL的深度学习计算框架,C++开发,支持多层感知器,长短时记忆模型,卷积神经网络,残差网络。调试方便,数据透明。无外部依赖。
License: Other
C++ 32.04%
C 67.85%
Makefile 0.11%
clnet's People
Contributors
clnet's Issues
Greetings,
When I try to run MLP example
sudo ./build/OpenCLNet /nf model MLP
I get an error:
[0,@2019-04-25 17:14:22] GeForce GTX 950M (kernels build: 87ms)
[debugger] interactive thread started on device 0.
[0,0,0ms] error rate: 0.303551
Current Tensor: clnet::type::FullyConnectedLayer: FCLayer_0=sigmoid(l0_weight*X+l0_bias)
Error in device 0: clEnqueueNDRangeKernel (-58): CL_INVALID_EVENT
Output from debugger:
[debugger] device 0 break on IterativeOptimizer: clnet::type::IterativeOptimizer
Greetings
Ratko
我在机器上运行了十多分钟没有出结果。Ctrl+C强制结束,然后机器就死机了。想问一下这可能是什么原因?
你好,我在vs2015+win10上编译成功,且命令行参数设置为MNIST_CNN /ds没有问题,成功训练手写体数字。但想只做前向传播命令行改为MNIST_CNN /p却没有什么结果,只显示:
[0,@1970-01-01 08:39:04] GeForce GTX 1060 6GB (kernels build: 280ms)
[debugger] interactive thread started on device 0.
[0] run time: 0ms.
请按任意键继续. . .
是工程还没有只检测的功能吗?还是在什么地方我还需要做一些修改?