Stuck on Queued

[ralghamdi]

I have installed TRAM locally, and have uploaded three different reports to be analyzed. It has been almost 2 hours for the newest one of them and they all still have the status of Queued as shown below:

I've been monitoring logs of the container with no indication of progress or errors.

[mehaase]

Hi @ralghamdi sorry for the delay in replying. When you run TRAM locally, you need to run this command in a separate tab:

tram pipeline run --model bert --run-forever &

That will run a background job that processes the reports. Please note that the BERT model is a bit slow when running on CPU (a few minutes per report) but the background process will display something in the logs when it starts processing a report.

[hAnguyen1517]

Hello @mehaase.
When you run the program and upload the documents, their statuses are in the queue:

But when you run the tram pipeline run --model bert --run-forever & in the new terminal, the files show the Error status.

Error from the terminal:

[mehaase]

Hi @hAnguyen1517. Are you running this locally? Did you download the BERT model? Step 9 in these instructions: https://github.com/center-for-threat-informed-defense/tram/wiki/Developers#developer-setup

[hAnguyen1517]

Hello @mehaase . Thank you so much for taking the time to respond to my question.

Yes, I am running tram locally using VS Code. I downloaded the BERT model and completed everything in step 9 as well. But it still won't work, and we are having issues loading and processing in the queue; I wonder if others are having the same issues.

[mehaase]

Can you please check that the following files exist in your installation and are approximately correct file size?

$ ls -lah data/ml-models/bert_model/
total 886352
drwxr-xr-x  5 mhaase  staff   160B Dec 19 15:34 ./
drwxr-xr-x  4 mhaase  staff   128B Aug 31  2023 ../
-rw-r--r--@ 1 mhaase  staff   395B Aug 31  2023 classes.txt
-rw-r--r--  1 mhaase  staff   2.7K Dec 19 15:34 config.json
-rw-r--r--  1 mhaase  staff   420M Dec 19 15:35 pytorch_model.bin