Training part of the dataset about paperrobot HOT 2 CLOSED

@AM-H Thank you very much for your interest in our research. Which part of the code did you run? The writing part or the link prediction part?

from paperrobot.

For paper reading, the data are loaded on LinkPredictionDataset, you can try to include part of the train2id.txt, valid2id.txt. Also, you can reduce the batch size by adding --batch_size 2 after python train.py.
For paper writing, the data are loaded on AssembleMem, you can try to change line 83 and line 86 of New paper writing/loader/preprocessing.py by truncating the data. You can reduce the batch size by adding --batch_size 2 after python train.py --data_path data/pubmed_abstract --model_dp abstract_model/.

from paperrobot.