Hello dear Singa Community,

I’m having a little trouble with creating a kCSVInput layer configuration.

Here is a part of my configuration:

...
layer {
  name: "data"
  type: kCSVInput
  store_conf {
    has_label: false
    backend: "textfile"
    path: "/root/data/entities.csv"
    batchsize: 100
  }
  include: kTrain
}
…

My CSV file looks like follows (no header line):

0,0,1,1,0,0,1,1,1,…long line...,0
0,0,1,1,0,0,1,1,1,…long line...,1
0,0,1,1,0,0,1,1,1,…long line...,1
0,0,1,1,0,0,0,0,1,…long line...,0

And so on.

After running with singa-run.sh I am getting the following exception:

F0710 18:49:50.371196  1224 csv.cc:63] Check failed: !t.Valid()
*** Check failure stack trace: ***
    @     0x7f47539ef9fd  google::LogMessage::Fail()
    @     0x7f47539f189d  google::LogMessage::SendToLog()
    @     0x7f47539ef5ec  google::LogMessage::Flush()
    @     0x7f47539f21be  google::LogMessageFatal::~LogMessageFatal()
    @     0x7f4753d20b47  singa::CSVInputLayer::Parse()
    @     0x7f4753d2efe3  singa::StoreInputLayer::ComputeFeature()
    @     0x7f4753d2f384  singa::SingleLabelRecordLayer::ComputeFeature()
    @     0x7f4753d007af  singa::CDWorker::TestOneBatch()
    @     0x7f4753d00c07  singa::Worker::Test()
    @     0x7f4753d0b3bc  singa::Worker::Run()
    @     0x7f4753791a60  (unknown)
    @     0x7f4752455184  start_thread
    @     0x7f47531ff37d  (unknown)


What am I missing? Am I using the correct input type and is my CSV correctly 
formatted?

Thank you for your help.


With best regards,


Georges

Reply via email to