EMNLP2018-JMEE icon indicating copy to clipboard operation
EMNLP2018-JMEE copied to clipboard

Published 20 hours ago • verified publisher icon lx865712528

How to set parameter "loss_alpha"

Open airkid opened this issue 6 years ago • 4 comments

Hi, I'm trying to run your code and I just found that value of parameter "loss_alpha" was not mentioned in your paper. Could you please give me a value to set this parameter? Thanks!

airkid avatar Feb 15 '19 02:02 airkid

I just found the code will crash if I set "loss_alpha">1.0.
Have you got the same problem?
@lx865712528

airkid avatar Feb 15 '19 03:02 airkid

Sorry about that. It is 5. Please refer to the new training script example, but I think before this update this parameter is hard-coded in this project.

lx865712528 avatar Feb 16 '19 14:02 lx865712528

Hi, I ran the newer update train script and found that the update rate for argument is still very low (F1<0.3) after many epochs.
I wonder if there is an misunderstanding.
I'd like to know the parameter "loss_alpha" and it is coded at
https://github.com/lx865712528/JMEE/blob/11b21e51dca943c7419c3a0e4d5f2f2e62acb1d1/enet/util.py#L189 And I guess maybe you mentioned "this parameter is hard-coded in this project" probably was form
https://github.com/lx865712528/JMEE/blob/494451d5852ba724d273ee6f97602c60a5517446/enet/run/ee/runner.py#L158

Could you please give me some advise about how to solve the argument F1 problems? THX！

airkid avatar Feb 19 '19 01:02 airkid

@airkid Have you solve this problem？

wangshuo6699 avatar Sep 03 '19 09:09 wangshuo6699