Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Error running DrQA

See original GitHub issue

I ran: python examples/drqa/train.py -t "#opensubtitles"

After 10 or 20 minutes:

05/25/2017 02:34:31 PM: [ Running validation... ]
Traceback (most recent call last):
  File "examples/drqa/train.py", line 180, in <module>
    main(opt)
  File "examples/drqa/train.py", line 118, in main
    valid_metric = validate(opt, doc_reader, iteration)
  File "examples/drqa/train.py", line 62, in validate
    valid_world.parley()
  File "/Users/bittlingmayer/Desktop/aca/ParlAI/parlai/core/worlds.py", line 221, in parley
    acts[1] = agents[1].act()
  File "/Users/bittlingmayer/Desktop/aca/ParlAI/parlai/agents/drqa/agents.py", line 157, in act
    [ex], null=self.word_dict['<NULL>'], cuda=self.opt['cuda']
  File "/Users/bittlingmayer/Desktop/aca/ParlAI/parlai/agents/drqa/utils.py", line 126, in batchify
    max_length = max([d.size(0) for d in docs])
  File "/Users/bittlingmayer/Desktop/aca/ParlAI/parlai/agents/drqa/utils.py", line 126, in <listcomp>
    max_length = max([d.size(0) for d in docs])
RuntimeError: dimension 0 out of range of 0D tensor at /Users/soumith/code/builder/wheel/pytorch-src/torch/lib/TH/generic/THTensor.c:24

Issue Analytics

State:
Created 6 years ago
Comments:8 (6 by maintainers)

Top GitHub Comments

1reaction

ajfischcommented, May 26, 2017

It has been tested on Squad and bAbI (the ones with answers in the context, e.g. tasks 1-5, 11-14, 16).

I’ll take a look and submit a fix whatever is going on here soon.

0reactions

jasewestoncommented, May 31, 2017

yes, a nice error message please!

On Wed, May 31, 2017 at 4:10 PM, Adam Fisch notifications@github.com wrote:

In my opinion, I think that it should be on the user to use the right tasks for the model (DrQA is not advertised to be a viable model for all of the tasks in ParlAI).

Therefore I think it is better to crash instead of silently skipping all of the examples. We could add a more descriptive error message…

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/facebookresearch/ParlAI/issues/101#issuecomment-305302326, or mute the thread https://github.com/notifications/unsubscribe-auth/AKjk-JOK2Fnqf2ks-Xa3SWJHu_poqVPYks5r_ckZgaJpZM4Nm7Dc .