Few Shot Dialogue Generation using DiKTNet
This repository reflects the final version for my Dialogue Generation Project during my internship at Saarthi.ai. The intermediate commits have been removed for privacy reasons.
The basic file structure of this repo is almost same as the original repository by @ishalyminov excluding some scripts.
utils/corpora.pyfor corpus creation and vocabulary generation of the DSTC Dataset.
utils/data_loaders.pyas the DataLoader for DSTC Dataset.
Data must be present in the main directory (For DSTC8 Dataset, folder named (dstc) is already present. Download the dataset, and unzip in here i.e. folders train, dev, test from dataset must be here).
The laed_features are added in the repository. These were generated using through the model trained during The given
laed_features folder has another folder named
features_dstc. This folder has the representations (specifically for DSTC8) generated using LAED architecture. They have been added, because, they will be used as the input to train.
Terminal Code for training FSDG
python train_fsdg.py \ DSTCZslCorpus \ --data_dir dstc/ \ --laed_z_folders laed_features/features_dstc/ \ --black_domains $domain \ --black_ratio 0.9 \ --action_match False \ --target_example_cnt 0 \ --random_seed $rnd `