Counterfactual Story Rewrite Based on Graph Transformer

Requires

• python>=3.6.2
• Tensorflow-gpu==1.14.0
• tensorflow-probability==0.7.0
• Texar-0.2.4 (this version has been included in this repository)
• xlnet_cased_L-12_H-768_A-12
• torch-1.3.0
• torchvision-0.3.0
• spacy==2.0.0
• nltk
• rouge
• edlib
• pandas
• boto3
• allennlp
• wmd (https://github.com/eaclark07/sms/tree/master/wmd-relax-master)
• matplotlib
• en_core_web_md-2.0.0.tar.gz
• pytorch-pretrained-BERT-master

---

Preprocessing

• jdk 1.8
• stanford-corenlp (http://nlp.stanford.edu/software/stanford-corenlp-full-2018-02-27.zip)

---

Training

• EXP1: python main_babymodel_1.1_lr.py
• EXP2: python main_trans512_auxiliary_01_00.py
• EXP3: python main_trans512_auxiliary_1_05.py
• EXP4: python main_2trans512_auxiliary_1_05.py
• EXP5: python main_xlnet.py
• EXP6: python main_babymodel_1.1_lr_SGT.py
• EXP7: python main_trans512_auxiliary_1_05_SGT.py

---

evaling

• EXP1: bash cmd_eval_babymodel1.1_lr.sh
• EXP2: bash cmd_eval_transformer512_auxiliary_loss_1_00.sh
• EXP3: bash cmd_eval_transformer512_auxiliary_loss_1_05.sh
• EXP4: bash cmd_eval_2transformer512_auxiliary_loss_1_05.sh
• EXP5: bash cmd_eval_xlnet.sh
• EXP6: bash cmd_eval_babymodel1.1_lr_SGT.sh
• EXP7: bash cmd_eval_transformer512_auxiliary_loss_1_05_SGT.sh

---

Result

Results in paper "Counterfactual Story Reasoning and Generation"

Results of Ours

TODO

• check all the traning code again and clean the code
• preprocessing for data

---

Acknowledge

• Counterfactual-StoryRW
• Texar-tf

第一次commit由于集群配置错误，导致账号不是本人的~