The code has been running for over 12 hours. Who can tell me if the code runs wrong? thank you.
Hello Zeyang. Does stdout provide any clues as to why the job has been running for so long? How long would it normally need?
Hi,
I ran the code again and now everything looks fine.It needs two days to run the test dataset. Who can tell me if the code is output correctly now?
i got this problem. i am a participator for the authorship verification task. i run train dataset is ok. the train dataset path is dir/pairs.jsonl. I think the test dataset path is the same as the train dataset, but i got this problem.
I have the exact same problem on the style change detection task.
I get the error message “No stdout recorded. Perhaps the software was never actually executed due to an initialization error in TIRA.” when running my software from the “My software” portal here on Tira.
I have verified that my software runs through and works perfectly fine on the vaildation (dev) set when copied to /home and the script ran from terminal inside the VM.
Seems to be a problem with Tira roughly after 40-45 minutes runtime… Have attempted to power on and off the VM a few times as well with no luck.
I have the same problem in Authorship Verification 2021 task. Until now my software can’t run on tira.
As I said in Why is no stdout recorded - #3 by danielembru I thing both post are related issues.
Every time I tried to run on the test I had in the output the message “No stdout recorded. Perhaps the software was never actually executed due to an initialization error in TIRA.”
I ran my software on training dataset 2 and everything went well for four hours (I am sure the software processed at least 6000 problems in this dataset) and suddenly the process ended and the outs also are “No stdout recorded. Perhaps the software was never actually executed due to an initialization error in TIRA.”
Could someone help me?
Thank you for your patience as we try to get to the bottom of this issue. TIRA currently fails to persist the output of a virtual machine when the virtual machine crashes. We are currently working on a bug fix that would solve this by getting the proper output of your software in case of crashing virtual machines.
While we are currently working on this bug fix, you may try to pinpoint the root of the problem by submitting “dummy runs,” that is, a run where a simple terminal command is submitted through the web interface instead of your actual software. Perhaps ls, cat, or touch are good first candidates, and if this succeeds, you can continue by executing more and more parts of your software.
Rest assured that this technical hurdle will be taken into account and the deadline adjusted accordingly.
Hello, until now I can’t run my software successfully in Tira. These are what I know:
- The software run inside the VM via ssh: Complete over a small dataset of 1000 problems; and run for 6 hours without troubles in the train dataset (I finished this process to continue trying directly in tira).
- (as I said before) In tira I ran my software on training dataset 2 and everything went well for four hours (I am sure the software processed at least 6000 problems in this dataset) and suddenly the process ended and the outs also are “No stdout recorded. Perhaps the software was never actually executed due to an initialization error in TIRA.”
- My software need about 1GB of RAM and the VM has 1 GB of free space to write the output
- Some executions over the test dataset fail as with the training dataset but two executions finished without this error (the executions were made without change in the VM and with exactly the same command in Tira).
- When I run the evaluator in these executions the process finished after a minute but the record says ‘deleted’. All the outs are hidden for me.
Because this is the first time I use the Tira system I am not sure what is the expected behavior, so all your comments will be helpful. Thank you
1 Like