stanfordnlp / treelstm

Tree-structured Long Short-Term Memory networks (http://arxiv.org/abs/1503.00075)
GNU General Public License v2.0
878 stars 237 forks source link

Segmentation fault (core dumped) Dependency Tree LSTM for Semantic Relatedness #8

Open aironashish opened 8 years ago

aironashish commented 8 years ago

Hi,

I am having the following error on dataset containing only two classes (two values of score- 1,2). Also, the segmentation fault error occurs on different epochs.
screen shot 2016-05-29 at 3 26 51 pm

I have made the following changes to handle two classes:

  1. self.num_classes = 2 and torch.range(1, 2):dot(output:exp()) in TreeLSTM.lua.
  2. dataset.labels[i] = (sim_file:readDouble() - 1) in read_data.lua.

I am running on Ubuntu 14.04.4 LTS and using LuaJIT 2.1.0-beta1.

StevenLOL commented 8 years ago

I made no change and yet have this problem occasionally.

th relatedness/main.lua --model dependency --layers 1 --dim 300 --epochs 10
sahith25 commented 7 years ago

Is there any update on this issue? I get segmentation fault randomly after some epochs

LiutongZhou commented 6 years ago

same issue encountered

collapsed at this line out = LSTM(latent_dim)(input)

error message: Segmentation fault (core dumped)