cudaBase.cpp CUDA基本操作
learnForest.cpp 是几年前写的,现不再维护,有需要时重构
nnet.h 神经网络头文件
nnetBase.cpp 神经网络配置解析
nnetConvolution.cpp 神经网络卷积层
nnetModel.cpp 神经网络训练+预测
optimization.h 优化算法头文件
optimLBFGS.cpp 优化算法LBFGS
optimSearch.cpp 优化算法步长搜索
optimVSGD.cpp 优化算法SGD
sparse.h 稀疏矩阵头文件
tensor.h 张量头文件
tensorVML.cpp 张量向量计算
xpu.h 设备头文件
imagenet224_hash_15.cfg 15层全卷积网络结构,256 bits 哈希学习
训练30轮准确率68.1%(single model single crop),相当于VGGNet 16层网络结构(4096 bits)的准确率
用GTX980ti单卡训练112小时,合每秒95帧
两卡训练加速1.8(最小的模型)~1.9+倍,测试发现对于并行加速,IO和带宽影响各占一半
-
Notifications
You must be signed in to change notification settings - Fork 0
hanjinze/nobodyDL
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
About
No description, website, or topics provided.
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published