search

cloudtrends / nlpbamboo

Automatically exported from code.google.com/p/nlpbamboo
GNU General Public License v3.0
0 stars 0 forks source link

It takes too long time to build the CRF model. #4

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
What steps will reproduce the problem?
1. I run the auto_build to build the CRF model, but it take over 72 hours.
It is not finished yet. But for CRF2 model, it just take 6 hourse.
2.
3.

What is the expected output? What do you see instead?

What version of the product are you using? On what operating system?

Please provide any additional information below.

Original issue reported on code.google.com by long....@gmail.com on 3 Nov 2008 at 1:55

GoogleCodeExporter commented 9 years ago 
Could you offer some your system information like cpu, mem, disk?

Original comment by det...@gmail.com on 10 Nov 2008 at 2:03

GoogleCodeExporter commented 9 years ago 
The Problem is that CRF++ needs lots of memory while training. If your box's 
starting
to use swap, it'll be very slow to finish the training. So, to build a crf 
model fast
it needs at least 1GB physical memory. That sounds horrible, but training is 
indeed a
memory bound process.

with my 1GB Thinkpad X60, it took about 12hour to finish the training.

Original comment by det...@gmail.com on 8 Dec 2008 at 12:54