Segmentation fault whilst training with transR

thunlp / Fast-TransX

An Efficient implementation of TransE and its extended models for Knowledge Representation Learning

MIT License

402 stars 108 forks source link

Segmentation fault whilst training with transR #37

Closed GittiHab closed 6 years ago

GittiHab commented 6 years ago

Hello, when running transR for training on the provided data I receive a segmentation fault. Could this be due to my machine or my parameters? Maybe you can provide an example script.

These are my parameters (which are more or less random to get it running):

./transR \
    -size 100 \
    -sizeR 100 \
    -input ../data/FB15K \
    -load ../data/FB15K \
    -output ./output \
    -load-binary 0 \
    -out-binary 0 \
    -thread  6 \
    -epochs 150 \
    -nbatches 1 \
    -alpha 0.1 \
    -margin 0.4 \
    -note ""

Thanks in advance!

ProKil commented 6 years ago

Sorry for our late response.

Segmentation faults most probably occur when input files are not at their right location. Could you please first check this issue?

Thanks, -Hao

Baschdl commented 6 years ago

You have to use a / at the end of the paths. It works with -input ../data/FB15K/ instead of -input ../data/FB15K

ProKil commented 6 years ago

Thanks @Baschdl very much. I will close this issue for now, and @GittiHab , if you still have questions, don't hesitate to reopen it.

-Hao