我无法在caffe中进行训练(ReadSolverParamsFromTextFileOrDie())

时间:2018-07-17 13:40:11

标签: caffe pycaffe matcaffe

我遇到以下错误:

I0717 22:26:05.513164 19438 upgrade_proto.cpp:1106] Attempting to upgrade input file specified using deprecated 'solver_type' field (enum)': solver.prototxt
I0717 22:26:05.513419 19438 upgrade_proto.cpp:1113] Successfully upgraded file specified using deprecated 'solver_type' field (enum) to 'type' field (string).
W0717 22:26:05.513458 19438 upgrade_proto.cpp:1115] Note that future Caffe releases will only support 'type' field (string) for a solver's type.
*** Aborted at 1531837565 (unix time) try "date -d @1531837565" if you are using GNU date ***
PC: @     0x7fe1148e8512 cfree
*** SIGSEGV (@0x7fff00000006) received by PID 19438 (TID 0x7fe1179c01c0) from PID 6; stack trace: ***
    @     0x7fe1148994b0 (unknown)
    @     0x7fe1148e8512 cfree
    @     0x7fe1161ad108 caffe::UpgradeSnapshotPrefixProperty()
    @     0x7fe1161ada60 caffe::ReadSolverParamsFromTextFileOrDie()
    @           0x41279a train()
    @           0x410a9f main
    @     0x7fe114884830 __libc_start_main
    @           0x4113c9 _start
    @                0x0 (unknown)
Segmentation fault (core dumped)

一旦我尝试使用Caffe训练模型,

这是求解器:

net: "trainval.prototxt"
base_lr:       0.0001
test_interval: 90
test_iter: 10
momentum:      0.99
lr_policy:     "step"
gamma:         0.1
weight_decay: 0.0005
max_iter:      300000
stepsize:      20000
display:       100
average_loss: 100
snapshot:      100000
snapshot_prefix: 'my_snapshot'
solver_mode:   GPU
solver_type: SGD
debug_info:    0

这是用于培训的bash文件:

#!/bin/bash
CAFFE_BIN=/home/user/caffe-models/caffe-opencl/build/tools/caffe

WEIGHTS=../snapshot_1.caffemodel # fine-tune from pretrained model

${CAFFE_BIN} train --solver=solver.prototxt -weights ${WEIGHTS} 2>&1 | tee ./stage1-`date +%F_%R`.log

能请你帮忙吗?我什至卸载了protobuf并再次安装。我找不到原因。

0 个答案:

没有答案