Segmentation fault (core dumped) Issue

[misbahmunir]

When I run the code on my data set (when I train it on more than 20 videos), it exits with the error - "Segmentation fault (core dumped) " at the validation stage.

There is no issue found at the level of Operating System or package installation at the moment.

[happsky]

@misbahmunir have you solved it? I meet the same error.

[happsky]

@garythung When I run the code on my data set (27 train videos, 18 val videos, 15 test videos), it exits with the error - "Segmentation fault (core dumped) " at the validation stage.

Here is the error:

19:41:46 Getting training data mean frame
19:41:50 Initializing LRCN
19:41:54 Starting training for 10 epochs
19:42:06 Epoch 1 training loss: 1.855981
19:42:17 Epoch 2 training loss: 1.719661
19:42:27 Epoch 3 training loss: 1.691099
19:42:27 Starting loss testing on the val split
Segmentation fault (core dumped)

[misbahmunir]

@happsky Yes, I did. I needed to upgrade the version of Lua. It would run smoothly with Lua 5.2.

[happsky]

@misbahmunir I have updated to Lua 5.2.3 via sudo apt-get install lua, but it still not work, can you tell me the way your updated?

[misbahmunir]

Here is the reference manual that I used: [http://torch.ch/docs/getting-started.html]