search

aksnzhy / xlearn

High performance, easy-to-use, and scalable machine learning (ML) package, including linear model (LR), factorization machines (FM), and field-aware factorization machines (FFM) for Python and CLI interface.
https://xlearn-doc.readthedocs.io/en/latest/index.html
Apache License 2.0
3.09k stars 519 forks source link

Segmentation fault while training FFM, but FM is work #184

Open ymkigeg opened 5 years ago

ymkigeg commented 5 years ago

First, I’m sorry about to post issue in chinese in last issue.

for this issue, I have transfer my data to libffm format， and traindata is 730M, validatedata is 290M my device info: 16cores + 64GB memory + 500G ssd

Thank you!

my log:

[------------] xLearn uses 4 threads for training task. [ ACTION ] Read Problem ... [------------] First check if the text file has been already converted to binary format. [------------] Binary file (../data/ffm/train_20181208.libffm.bin) NOT found. Convert text file to binary file. [------------] First check if the text file has been already converted to binary format. [------------] Binary file (../data/ffm/test_20181208.libffm.bin) NOT found. Convert text file to binary file. [------------] Number of Feature: 554365 [------------] Number of Field: 86 [------------] Time cost for reading problem: 24.53 (sec) [ ACTION ] Initialize model ... Segmentation fault

ymkigeg commented 5 years ago

when I set k=10，there is no problem any more above problem exists when k=50

aksnzhy commented 5 years ago

@ymkigeg Can you try to set k = 52 ? I just want to confirm the SSE instruction works well.

ymkigeg commented 5 years ago

I have tried below k values: k = 32, 36, 40, 41, 42, 43, 44 ok k = 45, 46, 47, 48, 50, 52, 100 fault