Question

我正在尝试训练一个像caffe train -solver solver.prototxt这样的网络。 Log说net.cpp:257] Network initialization done.，但后来崩溃了：

layer {
  name: "loss"
  type: "SoftmaxWithLoss"
  bottom: "fc8"
  bottom: "label"
  top: "loss"
}
I0206 21:22:21.640692  9495 layer_factory.hpp:77] Creating layer data
F0206 21:22:21.640879  9495 db_lmdb.hpp:15] Check failed: mdb_status == 0 (12 vs. 0) Cannot allocate memory
*** Check failure stack trace: ***
    @ 0xb7bb62b5  google::LogMessage::Fail()
    @ 0xb7bb857f  google::LogMessage::SendToLog()
    @ 0xb7bb5dee  google::LogMessage::Flush()
    @ 0xb7bb906d  google::LogMessageFatal::~LogMessageFatal()
    @ 0xb7d46bbf  caffe::db::LMDB::Open()
    @ 0xb7e7aadc  caffe::DataLayer<>::DataLayer()
    @ 0xb7e7ac6e  caffe::Creator_DataLayer<>()
    @ 0xb7e9bc86  caffe::LayerRegistry<>::CreateLayer()
    @ 0xb7edf788  caffe::Net<>::Init()
    @ 0xb7ee2439  caffe::Net<>::Net()
    @ 0xb7d32550  caffe::Solver<>::InitTestNets()
    @ 0xb7d33020  caffe::Solver<>::Init()
    @ 0xb7d33346  caffe::Solver<>::Solver()
    @ 0xb7d86c0b  caffe::Creator_SGDSolver<>()
    @  0x80586be  caffe::SolverRegistry<>::CreateSolver()
    @  0x8050fef  train()
    @  0x804df28  main
    @ 0xb734b637  __libc_start_main
    @  0x804e7fb  (unknown)

我找到了关于here这样的问题的讨论，但是solution对我不起作用，因为我没有与LMDB_MAP_SIZE中的/src/caffe/util/db_lmdb.cpp类似的东西

Caffe训练错误：检查失败：mdb_status == 0（12与0）无法分配内存

0 个答案: