VOCContextDataset and FCN8s classes aren't imported / defined in the Duration module. Also, within the VOCContextDataset, two parameters are undefined: img_size and transforms_train. Am I missing it?

During inference, if a sentence does not produce probabilities > 0.5 for any candidate event, the TempRelAPI.pred() pipeline breaks.

Specifically if the below line in construct_relations() returns an empty list, the TempRel pipeline breaks at a downstream line

# select event based on prob > 0.5, but eliminate ent_pred > context length
        ent_locs = [[x for x in (ent_probs[b,:, 1] > 0.5).nonzero().view(-1).tolist()
                     if x < lengths[b]] for b in range(batch_size)]

If ent_locs is an empty list, TempRel EventEvaluator.evaluate() call breaks at:

predicted_rels = [self.model._id_to_label[ridx] for ridx in torch.argmax(preds, dim=1).tolist()]

Also, NNClassifier.predict() doesn't like when probs returns as an empty list, either, which it definitely does return as an empty list if the ent_locs in construct_relations is an empty list, as rel_idxs returns as an empty list, etc:

probs = torch.cat(probs,dim=0)
labels = torch.cat(labels,dim=0)

I'm trying to find out how the ACE 2005 data gets written to the following pickle files which get loaded in the training pipeline of the BETTER component of this codebase. I do have a license for ACE 2005, but can't identify how it gets encoded for the training task.

p.add_argument('-train_pkl', type=str, default='./all_liz/train_w-pairs_xlmroberta.pkl')
p.add_argument('-dev_pkl', type=str, default='./all_liz/dev_w-pairs_xlmroberta.pkl')
p.add_argument('-test_pkl', type=str, default='./all_liz/test_w-pairs_xlmroberta.pkl')

Hi, I was following you guide but I got stuck on conda env create -f env.yml

The output says

Collecting package metadata (repodata.json): done
Solving environment: failed

ResolvePackageNotFound: 
  - gurobi==9.0.1=py37_0

I tried tinkering a bit with packets and versions but with no luck, any idea on how to solve this?

(Also, I tried for several days to reach your demo at https://kairos-event.isi.edu/ but apparently the site is down)

Thanks a lot

The TempRelAPI class references a matres_pipeline_best_hid90_dropout0.4_ew15.0.pth.tar file for the entity model. Is this checkpoint shared anywhere in particular?

You also mention https://github.com/rujunhan/EMNLP-2019 codebase for training. Which model type did you guys use to train the above checkpoint:

1. Single task
2. Multitask
3. Joint
4. Global

Thanks for any insight!

File "main.py", line 50, in response_pred
json_list = durationAPI.pred(events)
File "../Duration/inference_api.py", line 84, in pred
outputs = compute_predictions(self.model, test_loader)
File "/home/tianhao/anaconda3/envs/event-pipeline/lib/python3.7/site-packages/torch/autograd/grad_mode.py", line 49, in decorate_no_grad
return func(*args, **kwargs)
File "../Duration/utils_duration.py", line 40, in compute_predictions
p1_dur, p2_dur, fine, rel = model(words, span, root)
File "/home/tianhao/anaconda3/envs/event-pipeline/lib/python3.7/site-packages/torch/nn/modules/module.py", line 532, in call
result = self.forward(*input, **kwargs)
File "/home/tianhao/relation_extraction/EventPlus/component/Duration/scripts/src/factslab/factslab/pytorch/temporalmodule.py", line 241, in forward
inputs, masks = self.elmo_class.batch_to_embeddings(structures)
File "/home/tianhao/anaconda3/envs/event-pipeline/lib/python3.7/site-packages/torch/nn/modules/module.py", line 576, in getattr
type(self).name, name))
AttributeError: 'Elmo' object has no attribute 'batch_to_embeddings'

hi, I am not able to install and run the package. Is there another way to use it? thanks!

@ihungalexhsu @sunjiao123sun @VioletPeng @derekmma

Is there any chance you guys could expose an env.yml file that is cross platform compatible? If you export your conda env with --no-builds and --from-history, it'll only show the dependencies you asked for instead of all the packages that also get included alongside the packages in the original conda recipe. Right now the env.yml exposed in this repo works for Linux but not Mac OS.

conda env export --no-builds --from-history