diff --git a/README.md b/README.md index 83e3c1dfbd140d5f947cf673ea21d5e71dd7100b..e7ebcf205c94d13683decd2bf77bcf720a0eb615 100644 --- a/README.md +++ b/README.md @@ -33,7 +33,7 @@ Supported datasets are: - MultiWOZ 2.4 (https://github.com/smartyfh/MultiWOZ2.4.git) - Unified data format (currently supported: MultiWOZ 2.1) (see https://github.com/ConvLab/ConvLab-3) -See the README file in `data/` for more details how to obtain and prepare the datasets for use in TripPy. +See the README file in 'data/' for more details how to obtain and prepare the datasets for use in TripPy. The ```--task_name``` is - 'sim-m', for sim-M @@ -41,6 +41,7 @@ The ```--task_name``` is - 'woz2', for WOZ 2.0 - 'multiwoz21', for MultiWOZ 2.0-2.4 - 'multiwoz21_legacy', for MultiWOZ 2.1 legacy version +- 'unified', for ConvLab-3's unified data format With a sequence length of 180, you should expect the following average JGA: - 53% for MultiWOZ 2.0 @@ -55,9 +56,9 @@ With a sequence length of 180, you should expect the following average JGA: ## ConvLab-3 -TripPy is integrated in ConvLab-3 as ready-to-use dialogue state tracker. A checkpoint is available at HuggingFace (see the ConvLab-3 repo for more details). +TripPy is integrated in ConvLab-3 as ready-to-use dialogue state tracker. A checkpoint is available at HuggingFace (see the [ConvLab-3 repo](https://github.com/ConvLab/ConvLab-3) for more details). -If you want to train your own TripPy model for ConvLab-3 from scratch, you can do so by using this code, setting ```--task_name='unified'```. The ```--data_dir``` parameter will be ignored in that case. Pick the file for ```--dataset_config``` according to the dataset you want to train for. For MultiWOZ, this would 'data/unified_multiwoz21'. +If you want to train your own TripPy model for ConvLab-3 from scratch, you can do so by using this code, setting ```--task_name='unified'```. The ```--data_dir``` parameter will be ignored in that case. Pick the file for ```--dataset_config``` according to the dataset you want to train for. For MultiWOZ, this would 'data/unified_multiwoz21.json'. ## Requirements @@ -85,7 +86,7 @@ If you use TripPy in your own work, please cite our work as follows: } ``` -This repository also contains the code of our paper [Out-of-Task Training for Dialog State Tracking Models"](https://www.aclweb.org/anthology/2020.coling-main.596). +This repository also contains the code of our paper [Out-of-Task Training for Dialog State Tracking Models](https://www.aclweb.org/anthology/2020.coling-main.596). If you use TripPy for MTL, please cite our work as follows: