From e9d2d39c9dde78ba0d15ed231b722663f701d53f Mon Sep 17 00:00:00 2001 From: lewtun Date: Mon, 10 May 2021 11:21:19 +0200 Subject: [PATCH] Add SubjQA dataset (#2302) * Add SubjQA dataset * Remove unused code * Add README * Add dataset infos and dummy data * Fix style * Update datasets/subjqa/README.md Co-authored-by: Quentin Lhoest <42851186+lhoestq@users.noreply.github.com> * Update datasets/subjqa/README.md Co-authored-by: Quentin Lhoest <42851186+lhoestq@users.noreply.github.com> * Refactor SubjQA to produce SQuAD schema * Simplify extraction of answer metadata * Remove redundant feature * Update dataset infos * Trim down size of dummy data * Update README * Fix field description in README * Add dataset statistics and collection info to the README * Add info on dataset creation * Add annotation process and social impact to README * Update datasets/subjqa/README.md Co-authored-by: Quentin Lhoest <42851186+lhoestq@users.noreply.github.com> * Update datasets/subjqa/README.md Co-authored-by: Quentin Lhoest <42851186+lhoestq@users.noreply.github.com> * Replace eval with ast.literal_eval for safety! * fix missing extended| prefix in source datasets tags Co-authored-by: Quentin Lhoest <42851186+lhoestq@users.noreply.github.com> --- datasets/subjqa/README.md | 288 ++++++++++++++++++ datasets/subjqa/dataset_infos.json | 1 + .../subjqa/dummy/books/1.1.0/dummy_data.zip | Bin 0 -> 12042 bytes .../dummy/electronics/1.1.0/dummy_data.zip | Bin 0 -> 12189 bytes .../subjqa/dummy/grocery/1.1.0/dummy_data.zip | Bin 0 -> 7371 bytes .../subjqa/dummy/movies/1.1.0/dummy_data.zip | Bin 0 -> 11486 bytes .../dummy/restaurants/1.1.0/dummy_data.zip | Bin 0 -> 9349 bytes .../dummy/tripadvisor/1.1.0/dummy_data.zip | Bin 0 -> 10747 bytes datasets/subjqa/subjqa.py | 211 +++++++++++++ 9 files changed, 500 insertions(+) create mode 100644 datasets/subjqa/README.md create mode 100644 datasets/subjqa/dataset_infos.json create mode 100644 datasets/subjqa/dummy/books/1.1.0/dummy_data.zip create mode 100644 datasets/subjqa/dummy/electronics/1.1.0/dummy_data.zip create mode 100644 datasets/subjqa/dummy/grocery/1.1.0/dummy_data.zip create mode 100644 datasets/subjqa/dummy/movies/1.1.0/dummy_data.zip create mode 100644 datasets/subjqa/dummy/restaurants/1.1.0/dummy_data.zip create mode 100644 datasets/subjqa/dummy/tripadvisor/1.1.0/dummy_data.zip create mode 100644 datasets/subjqa/subjqa.py diff --git a/datasets/subjqa/README.md b/datasets/subjqa/README.md new file mode 100644 index 00000000000..5c896afb47a --- /dev/null +++ b/datasets/subjqa/README.md @@ -0,0 +1,288 @@ +--- +annotations_creators: +- expert-generated +language_creators: +- found +languages: +- en +licenses: +- unknown +multilinguality: +- monolingual +size_categories: +- 1K The platform provides quality control by showing the workers 5 questions at a time, out of which one is labeled by the experts. A worker who fails to maintain 70% accuracy is kicked out by the platform and his judgements are ignored ... To ensure good quality labels, we paid each worker 5 cents per annotation. + +The instructions for generating a question are shown in the following figure: + +ques_gen + +Similarly, the interface for the answer span and subjectivity labelling tasks is shown below: + +![span_collection](https://user-images.githubusercontent.com/26859204/117259223-1fda1480-ae4e-11eb-9305-658ee6e3971d.png) + +As described in the SubjQA paper, the workers assign subjectivity scores (1-5) to each question and the selected answer span. They can also indicate if a question cannot be answered from the given review. + + +#### Who are the annotators? + +Workers on the Appen platform. + +### Personal and Sensitive Information + +[Needs More Information] + +## Considerations for Using the Data + +### Social Impact of Dataset + +The SubjQA dataset can be used to develop question-answering systems that can provide better on-demand answers to e-commerce customers who are interested in subjective questions about products and services. + +### Discussion of Biases + +[Needs More Information] + +### Other Known Limitations + +[Needs More Information] + +## Additional Information + +### Dataset Curators + +The people involved in creating the SubjQA dataset are the authors of the accompanying paper: + +* Johannes Bjerva1, Department of Computer Science, University of Copenhagen, Department of Computer Science, Aalborg University +* Nikita Bhutani, Megagon Labs, Mountain View +* Behzad Golshan, Megagon Labs, Mountain View +* Wang-Chiew Tan, Megagon Labs, Mountain View +* Isabelle Augenstein, Department of Computer Science, University of Copenhagen + +### Licensing Information + +The SubjQA dataset is provided "as-is", and its creators make no representation as to its accuracy. + +The SubjQA dataset is constructed based on the following datasets and thus contains subsets of their data: +* [Amazon Review Dataset](http://jmcauley.ucsd.edu/data/amazon/links.html) from UCSD + * Used for _books_, _movies_, _grocery_, and _electronics_ domains +* [The TripAdvisor Dataset](http://times.cs.uiuc.edu/~wang296/Data/) from UIUC's Database and Information Systems Laboratory + * Used for the _TripAdvisor_ domain +* [The Yelp Dataset](https://www.yelp.com/dataset) + * Used for the _restaurants_ domain + +Consequently, the data within each domain of the SubjQA dataset should be considered under the same license as the dataset it was built upon. + +### Citation Information + +If you are using the dataset, please cite the following in your work: +``` +@inproceedings{bjerva20subjqa, + title = "SubjQA: A Dataset for Subjectivity and Review Comprehension", + author = "Bjerva, Johannes and + Bhutani, Nikita and + Golahn, Behzad and + Tan, Wang-Chiew and + Augenstein, Isabelle", + booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing", + month = November, + year = "2020", + publisher = "Association for Computational Linguistics", +} +``` + +### Contributions + +Thanks to [@lewtun](https://github.com/lewtun) for adding this dataset. diff --git a/datasets/subjqa/dataset_infos.json b/datasets/subjqa/dataset_infos.json new file mode 100644 index 00000000000..8ef13ca17bb --- /dev/null +++ b/datasets/subjqa/dataset_infos.json @@ -0,0 +1 @@ +{"books": {"description": "SubjQA is a question answering dataset that focuses on subjective questions and answers.\nThe dataset consists of roughly 10,000 questions over reviews from 6 different domains: books, movies, grocery,\nelectronics, TripAdvisor (i.e. hotels), and restaurants.", "citation": "@inproceedings{bjerva20subjqa,\n title = \"SubjQA: A Dataset for Subjectivity and Review Comprehension\",\n author = \"Bjerva, Johannes and\n Bhutani, Nikita and\n Golahn, Behzad and\n Tan, Wang-Chiew and\n Augenstein, Isabelle\",\n booktitle = \"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing\",\n month = November,\n year = \"2020\",\n publisher = \"Association for Computational Linguistics\",\n}\n", "homepage": "", "license": "", "features": {"domain": {"dtype": "string", "id": null, "_type": "Value"}, "nn_mod": {"dtype": "string", "id": null, "_type": "Value"}, "nn_asp": {"dtype": "string", "id": null, "_type": "Value"}, "query_mod": {"dtype": "string", "id": null, "_type": "Value"}, "query_asp": {"dtype": "string", "id": null, "_type": "Value"}, "q_reviews_id": {"dtype": "string", "id": null, "_type": "Value"}, "question_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ques_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ques_subjective": {"dtype": "bool", "id": null, "_type": "Value"}, "review_id": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "answer_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ans_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ans_subjective": {"dtype": "bool", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "subjqa", "config_name": "books", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 2473128, "num_examples": 1314, "dataset_name": "subjqa"}, "test": {"name": "test", "num_bytes": 649413, "num_examples": 345, "dataset_name": "subjqa"}, "validation": {"name": "validation", "num_bytes": 460214, "num_examples": 256, "dataset_name": "subjqa"}}, "download_checksums": {"https://github.com/lewtun/SubjQA/archive/refs/heads/master.zip": {"num_bytes": 11384657, "checksum": "f3d58fd04c698fccb326b7ea4ea93098cc2186a3925f4bbad9b538ed7acc72db"}}, "download_size": 11384657, "post_processing_size": null, "dataset_size": 3582755, "size_in_bytes": 14967412}, "electronics": {"description": "SubjQA is a question answering dataset that focuses on subjective questions and answers.\nThe dataset consists of roughly 10,000 questions over reviews from 6 different domains: books, movies, grocery,\nelectronics, TripAdvisor (i.e. hotels), and restaurants.", "citation": "@inproceedings{bjerva20subjqa,\n title = \"SubjQA: A Dataset for Subjectivity and Review Comprehension\",\n author = \"Bjerva, Johannes and\n Bhutani, Nikita and\n Golahn, Behzad and\n Tan, Wang-Chiew and\n Augenstein, Isabelle\",\n booktitle = \"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing\",\n month = November,\n year = \"2020\",\n publisher = \"Association for Computational Linguistics\",\n}\n", "homepage": "", "license": "", "features": {"domain": {"dtype": "string", "id": null, "_type": "Value"}, "nn_mod": {"dtype": "string", "id": null, "_type": "Value"}, "nn_asp": {"dtype": "string", "id": null, "_type": "Value"}, "query_mod": {"dtype": "string", "id": null, "_type": "Value"}, "query_asp": {"dtype": "string", "id": null, "_type": "Value"}, "q_reviews_id": {"dtype": "string", "id": null, "_type": "Value"}, "question_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ques_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ques_subjective": {"dtype": "bool", "id": null, "_type": "Value"}, "review_id": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "answer_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ans_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ans_subjective": {"dtype": "bool", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "subjqa", "config_name": "electronics", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 2123648, "num_examples": 1295, "dataset_name": "subjqa"}, "test": {"name": "test", "num_bytes": 608899, "num_examples": 358, "dataset_name": "subjqa"}, "validation": {"name": "validation", "num_bytes": 419042, "num_examples": 255, "dataset_name": "subjqa"}}, "download_checksums": {"https://github.com/lewtun/SubjQA/archive/refs/heads/master.zip": {"num_bytes": 11384657, "checksum": "f3d58fd04c698fccb326b7ea4ea93098cc2186a3925f4bbad9b538ed7acc72db"}}, "download_size": 11384657, "post_processing_size": null, "dataset_size": 3151589, "size_in_bytes": 14536246}, "grocery": {"description": "SubjQA is a question answering dataset that focuses on subjective questions and answers.\nThe dataset consists of roughly 10,000 questions over reviews from 6 different domains: books, movies, grocery,\nelectronics, TripAdvisor (i.e. hotels), and restaurants.", "citation": "@inproceedings{bjerva20subjqa,\n title = \"SubjQA: A Dataset for Subjectivity and Review Comprehension\",\n author = \"Bjerva, Johannes and\n Bhutani, Nikita and\n Golahn, Behzad and\n Tan, Wang-Chiew and\n Augenstein, Isabelle\",\n booktitle = \"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing\",\n month = November,\n year = \"2020\",\n publisher = \"Association for Computational Linguistics\",\n}\n", "homepage": "", "license": "", "features": {"domain": {"dtype": "string", "id": null, "_type": "Value"}, "nn_mod": {"dtype": "string", "id": null, "_type": "Value"}, "nn_asp": {"dtype": "string", "id": null, "_type": "Value"}, "query_mod": {"dtype": "string", "id": null, "_type": "Value"}, "query_asp": {"dtype": "string", "id": null, "_type": "Value"}, "q_reviews_id": {"dtype": "string", "id": null, "_type": "Value"}, "question_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ques_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ques_subjective": {"dtype": "bool", "id": null, "_type": "Value"}, "review_id": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "answer_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ans_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ans_subjective": {"dtype": "bool", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "subjqa", "config_name": "grocery", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1317488, "num_examples": 1124, "dataset_name": "subjqa"}, "test": {"name": "test", "num_bytes": 721827, "num_examples": 591, "dataset_name": "subjqa"}, "validation": {"name": "validation", "num_bytes": 254432, "num_examples": 218, "dataset_name": "subjqa"}}, "download_checksums": {"https://github.com/lewtun/SubjQA/archive/refs/heads/master.zip": {"num_bytes": 11384657, "checksum": "f3d58fd04c698fccb326b7ea4ea93098cc2186a3925f4bbad9b538ed7acc72db"}}, "download_size": 11384657, "post_processing_size": null, "dataset_size": 2293747, "size_in_bytes": 13678404}, "movies": {"description": "SubjQA is a question answering dataset that focuses on subjective questions and answers.\nThe dataset consists of roughly 10,000 questions over reviews from 6 different domains: books, movies, grocery,\nelectronics, TripAdvisor (i.e. hotels), and restaurants.", "citation": "@inproceedings{bjerva20subjqa,\n title = \"SubjQA: A Dataset for Subjectivity and Review Comprehension\",\n author = \"Bjerva, Johannes and\n Bhutani, Nikita and\n Golahn, Behzad and\n Tan, Wang-Chiew and\n Augenstein, Isabelle\",\n booktitle = \"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing\",\n month = November,\n year = \"2020\",\n publisher = \"Association for Computational Linguistics\",\n}\n", "homepage": "", "license": "", "features": {"domain": {"dtype": "string", "id": null, "_type": "Value"}, "nn_mod": {"dtype": "string", "id": null, "_type": "Value"}, "nn_asp": {"dtype": "string", "id": null, "_type": "Value"}, "query_mod": {"dtype": "string", "id": null, "_type": "Value"}, "query_asp": {"dtype": "string", "id": null, "_type": "Value"}, "q_reviews_id": {"dtype": "string", "id": null, "_type": "Value"}, "question_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ques_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ques_subjective": {"dtype": "bool", "id": null, "_type": "Value"}, "review_id": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "answer_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ans_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ans_subjective": {"dtype": "bool", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "subjqa", "config_name": "movies", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 2986348, "num_examples": 1369, "dataset_name": "subjqa"}, "test": {"name": "test", "num_bytes": 620513, "num_examples": 291, "dataset_name": "subjqa"}, "validation": {"name": "validation", "num_bytes": 589663, "num_examples": 261, "dataset_name": "subjqa"}}, "download_checksums": {"https://github.com/lewtun/SubjQA/archive/refs/heads/master.zip": {"num_bytes": 11384657, "checksum": "f3d58fd04c698fccb326b7ea4ea93098cc2186a3925f4bbad9b538ed7acc72db"}}, "download_size": 11384657, "post_processing_size": null, "dataset_size": 4196524, "size_in_bytes": 15581181}, "restaurants": {"description": "SubjQA is a question answering dataset that focuses on subjective questions and answers.\nThe dataset consists of roughly 10,000 questions over reviews from 6 different domains: books, movies, grocery,\nelectronics, TripAdvisor (i.e. hotels), and restaurants.", "citation": "@inproceedings{bjerva20subjqa,\n title = \"SubjQA: A Dataset for Subjectivity and Review Comprehension\",\n author = \"Bjerva, Johannes and\n Bhutani, Nikita and\n Golahn, Behzad and\n Tan, Wang-Chiew and\n Augenstein, Isabelle\",\n booktitle = \"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing\",\n month = November,\n year = \"2020\",\n publisher = \"Association for Computational Linguistics\",\n}\n", "homepage": "", "license": "", "features": {"domain": {"dtype": "string", "id": null, "_type": "Value"}, "nn_mod": {"dtype": "string", "id": null, "_type": "Value"}, "nn_asp": {"dtype": "string", "id": null, "_type": "Value"}, "query_mod": {"dtype": "string", "id": null, "_type": "Value"}, "query_asp": {"dtype": "string", "id": null, "_type": "Value"}, "q_reviews_id": {"dtype": "string", "id": null, "_type": "Value"}, "question_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ques_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ques_subjective": {"dtype": "bool", "id": null, "_type": "Value"}, "review_id": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "answer_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ans_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ans_subjective": {"dtype": "bool", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "subjqa", "config_name": "restaurants", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1823331, "num_examples": 1400, "dataset_name": "subjqa"}, "test": {"name": "test", "num_bytes": 335453, "num_examples": 266, "dataset_name": "subjqa"}, "validation": {"name": "validation", "num_bytes": 349354, "num_examples": 267, "dataset_name": "subjqa"}}, "download_checksums": {"https://github.com/lewtun/SubjQA/archive/refs/heads/master.zip": {"num_bytes": 11384657, "checksum": "f3d58fd04c698fccb326b7ea4ea93098cc2186a3925f4bbad9b538ed7acc72db"}}, "download_size": 11384657, "post_processing_size": null, "dataset_size": 2508138, "size_in_bytes": 13892795}, "tripadvisor": {"description": "SubjQA is a question answering dataset that focuses on subjective questions and answers.\nThe dataset consists of roughly 10,000 questions over reviews from 6 different domains: books, movies, grocery,\nelectronics, TripAdvisor (i.e. hotels), and restaurants.", "citation": "@inproceedings{bjerva20subjqa,\n title = \"SubjQA: A Dataset for Subjectivity and Review Comprehension\",\n author = \"Bjerva, Johannes and\n Bhutani, Nikita and\n Golahn, Behzad and\n Tan, Wang-Chiew and\n Augenstein, Isabelle\",\n booktitle = \"Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing\",\n month = November,\n year = \"2020\",\n publisher = \"Association for Computational Linguistics\",\n}\n", "homepage": "", "license": "", "features": {"domain": {"dtype": "string", "id": null, "_type": "Value"}, "nn_mod": {"dtype": "string", "id": null, "_type": "Value"}, "nn_asp": {"dtype": "string", "id": null, "_type": "Value"}, "query_mod": {"dtype": "string", "id": null, "_type": "Value"}, "query_asp": {"dtype": "string", "id": null, "_type": "Value"}, "q_reviews_id": {"dtype": "string", "id": null, "_type": "Value"}, "question_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ques_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ques_subjective": {"dtype": "bool", "id": null, "_type": "Value"}, "review_id": {"dtype": "string", "id": null, "_type": "Value"}, "id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "answer_subj_level": {"dtype": "int64", "id": null, "_type": "Value"}, "ans_subj_score": {"dtype": "float32", "id": null, "_type": "Value"}, "is_ans_subjective": {"dtype": "bool", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "builder_name": "subjqa", "config_name": "tripadvisor", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1575021, "num_examples": 1165, "dataset_name": "subjqa"}, "test": {"name": "test", "num_bytes": 689508, "num_examples": 512, "dataset_name": "subjqa"}, "validation": {"name": "validation", "num_bytes": 312645, "num_examples": 230, "dataset_name": "subjqa"}}, "download_checksums": {"https://github.com/lewtun/SubjQA/archive/refs/heads/master.zip": {"num_bytes": 11384657, "checksum": "f3d58fd04c698fccb326b7ea4ea93098cc2186a3925f4bbad9b538ed7acc72db"}}, "download_size": 11384657, "post_processing_size": null, "dataset_size": 2577174, "size_in_bytes": 13961831}} \ No newline at end of file diff --git a/datasets/subjqa/dummy/books/1.1.0/dummy_data.zip b/datasets/subjqa/dummy/books/1.1.0/dummy_data.zip new file mode 100644 index 0000000000000000000000000000000000000000..6135c4d1f6900148da1d010f2ef4e8afa156dc9c GIT binary patch literal 12042 zcmbuF1yCeSwzhG%!QI{6VQ_c%!QE+`!QE-x-Q5OvhXDq6cXt?U_~%=>EC1b%8@s2Y zyR$kf-#1TIM0K95M@beO0v+U!$h0d~`BURBAMn4wO=@+SqjKMm-|KMju#1pn`c_jj)UsPN|_CUrGf z5b)Fxpy3~HR}Xj)P>2Um5RgCR{;yMG`9BWxFDAwg4mK|TCTs5hM9#(0)(Y@9RhU0r zDE(dxVi+V4kcZ!rLjQ#-z{~}}XyW2_q1^{uYyH%9M5pB9q#|BPJ}Jef%2Q-}<+apo z!)N9_(bgt}ioQbv3kf2*xOLP8C1tfZ#Sn(yq<)q_jbaJt@qNdwo4NmDt3O|}YI}YUEie&&Q!mNgE^|s~7zJ2}_YC`?SwKuqMwBvKEB$RMWFhN<0p5 zc-`OQRDM&T|2vVYuvaD<)su+h@ssyuse^K2P2TJbD2xjBya9oqZ}rGefzXxuSx?$V`K>%g_NA~z8`E`Lw|4vGN&M!e6DU3Kn` zUrZX&qIN#vS}^hUaawo*IZGujUTA@v5hjmQlC^OG-v$zmyDy&Akd+ONDKdOwY*|g)U}})$arTD{}=PLE;&6pn!5bfko_&y7;Ui#f&-ruik;=m4(*$dZSBz zAdMkjN7zLKTgyhsu-c^{N!c~CImp#9loy+>53uCk7WL9;y|<}^9|!}`589H!5K8r1 z1uO$Dt`Io1-LfCu8VwN36nE3mB6ElS;S zo`PET@w7pA5YjrKHSi$QjGY-VcD0ChFpxC~D2Bc(MUh>%@v*9#V&bej@ya_K`zw0e z^pfn7A=vD^-w`POIPT44_hGGbHztk&VsyunfUUwV=%_nJ)zHdu`rjL5$gl}dZ*JKf zQkX0Q3UMq!HCP5T-Q|dLq%xw36E$lj-QZs%>hCx!6oTgHJOXCKkvhG0eM9hi|@ns^Ah%Qj&yyZDw;n6>#@uPDXSq3xMidTfn=l8YZwiETi%|~5aKi1!7tu)L z4g~AGc^Xy{@gR{`K`WUdRz0TS7!?Pe#L3aX2bX6*WLIeg$g5C-XmcvKVU!cHydLL1 z5ht{09aq=*cO%10Ir`(1u6d}sn$)=Tw?q0rpw>O-olW5z> zQ+Zb(^qYfGR|>_gbJ37^FR{^Ka|bPPf23dWfTdwCM^Om4k=IS`hA7^P1(+v@T3yjJ zN8;S87-+jTDT<4-GY&FrkKhAMM>$Yaltd>CZW9M3dJSmRRR=fU?%ow)L+uJ0`r-#$ zCo?2mA|(a1TFIFSLDAr^R4gGrIn@mf2a^X=9O@5XkjDrV*j0}HI6QTjg+rTFv2UdZ zLozgQ?`7>}A#RZxFjdAKX$LjHMj4EmT1b+P|Q7)pL%kD2ozwU zl*(3>loK%D^u;tIy^ID=m&BgaO?kzEA$4ITnMcy#XaH-l8Fe1fC6h>og2UD!VMARP zW)FYEv@N8L+)RVV6eA2iI?}@zrHyVb)JG2DGfj4r!;O!ZnOJPo2g1Sci3v#M#g^(T zQY_8lVPKCwCm~8z3J96v@I>NIuEVMsd)&cRl>(}>VVk7sQQT3cctT-ONS)c&lRO-* zkcj2#5LOWhRB4`>Pd{bxNjs*~U=@6Zx}S*f7bmNm zrTFe;J(oYlOS=&92@!g6d{*Lf$o<^RENNe$-r~$eJ59knX z=S8sV0po(?EKHhiOd%lB`Gay06OMH#hd_)5A(G;>yoPEu)>uKDh50)_g#)&Cid)1n z3d_uGG#A)BoXuE`rmIjE_l~*b~AP zOgzam1DTJ?U^JbIZ<0-xY_uj-9xzLJqNZpWIK+2@N4@VEg$v3$R4LD&$9#FB0MQ_S+w8xlDsje zY;(2=t55gB^Z5W$A!iHJdEe#I?r`Z6GEj@u%Nf5g06lR-M0nD3w*J+s9Jrkk z$wyqhEE!1EZnae7LDPoz7G`sYzu;Qx%}x!9jXh{Hf*!XMq*2#yOWT(YN^_jx9 z6S-s_{j_m?IRo?$7Z$E8Cu(ZYfBD~0V)(eBpzD}h_7~Q7+MB>exFn^_&j>f*Z6ft+{bh2D49e%~8WO=%Oz z{<3BChT6Pe`TI!1w^2jXM6jsM<$6$wqWn>lhMqE?9cZ(o66EutJk{~Y0}bH7Dy0Lf z0Wd0&rGabhU7lX@dD@>neR}8c(OKrg+bu3q_CIb`o*WF3R8LvoP6|#=Uo%TC%)jcF zT^znVIxhj*Y}(g3^e;0Q-Q`m|N+$H@4r)&RgP&iv);Zpx|5^(Fxyn=+$5?Xuy$>+- zyTt$eUsRcspq~8R6V|2OP+eZADbNcR^OOs_r zqCd7eA{3XR(0SraaA01-t$kI4MYOm6_E=vkd1}V6F|r6fZu{n;U*Ogdmn0A&ueoAG z$bRqWVKcau>3tS$M`PY9sN5yxO7N#zJNl&C;svVtng_yfTKFX3d(s zYL%2jhtHZ>Z5QvWHo-j^O_`sD=8p2o@N{AZl9x)=L}k_JZ3peE@KZvAezk0timirv zGlNyfxQ)KLMD1B~Ial$Nfe&~5ORD=%6VW)d0$$HjBsV`LlOLSG(?B9MymM3Cti$Ne zc@&=F1@P?oT0#5;Xg-wMVH%Y!)zP}O&pIudo6SY{gyWPHiR2q}8{e3Xl!g_d58fc0 zysu)kKj^{y?Z&|d)8k2KrTRI1_q_8%3mUpr&^YU%D!f>a{fnp;00S2WDko0;3 z{aHk9x3?{jLd8*F`U03%2^)Z8&=WN?ccvXUr7N+6Vc;{td}*_uRb+V<36p^wS*=tW zg^Kq4-|Kh6I0^Wnxb=p3Vt-s;QU=-)BhGEh4Z|4sWl9zl296m#J~D$zxLNahc0SDh zkU}z-i&O=}oa~J?j_^1F2;HZODhdut6Wm~a#@TBV_{rz$(mJ?)eG(g2*vmtI1g)IZ z1r0znr9TTRI}vyjd?8F9?a*a)$*aVY0wWF+4s>sCnZp6%hy@K*Te}u{j1UjkyIl>^ zuk%btAaIGmmWoF_=1Hy(L0Ce8=dpVok7cIOt1$bM@cW1EWYr(+KYHc*o7=DRiMVo1 z_T7+G!C{`Dp*Kgv{5&u)Wb4a>Armpx7?rk3xC8N!Q*~Cx zMccpWgbJw;&5vsctGBPtzsdrMVGViAyW?S)Ezqwv2&H9Ud&7?g8%k;CEx|GJMbTsV zyDi$iYtM6VUru7(T|jo%{sF)cSO_xV+f9z??O?cnjDn)K3H^ZAuYlUDQ{?wq5``yT zjL|6d0gT-zj(`aVWp~0kv2$ll z7@WqaaPRY}^B-j11SPz44 z;4sWh`He^O0HJN8UHqf>2pE4(1M*NdF_Vau}`&e(xfuDmkaKSVLZ_ z4vS{$`fz20_eAmO7s3}yc-NCGY#gWR1{e33nYS1A9lB?MZpoYJ#`?#UbX-l0h_1zB z?z%}lt8R9>Y9+Q@C3F?E-g}Bw0X`9GTcXaL2Q>tthCR?5$2JVJC-MH z4!rt8339`Q^H{Xp{dCt*Zm_uInUu@|d)}JT+nc^!6ab0Vn-B|jm3)&dCVohaL%`J# zf9I5}sc@<@pQfPwk@|)9A%k6a2Csy~&rl|iBK7@8P}`Yi6Q0wpz#NK{d%w6z=}-YU zEgn`BRS=FbN=WhA-Lxi&iU5+2+5NVd$yDW9s$CML5TJV;G5CU#U5R?LN zCN5=DWegLO-9dyu>IETZitfH{Hzk(zlh(uKFLjbz5Go@l=dP>pA!5K4Fr+pMHmpj0 z8_Hq6LmNu!(?>YPn*$@1$rgHascCqTay>ZA?^EyV?hIA}*r$^g^?+&yeX?jr5x->~ zn1Ln>H%TYGCg)D^-pl^g024dsOm$V=C zF!nj?NJd+R%IxZ&NPzZY;J0s^cTL|t7^uw?)L?+82nk2AK2i5?>9D8^M*=_XSe}SJ zmU6wWMuLhz-yG-8LndD`QzB4n|*{Vi(6T_}b#$SjzHrvAJb6X-Q>$#U&xdoeLp z@HmX~6@yHKcTml}SV=_76i{$4vy^eWpa8rUVU!&S{hBcnu%;3Gj26R~sz9fnnTa!; z2Ox>IxS=YYAC!Z@E4GkO?N>z?KdFtU*@$*aL{;Zu*VB19iC7J4>?t*u*7`mzubAul zEj1SRf~IBuo#)|bZj6^3;E!-&i!TPXL<_*K6JmufUf@k5-k8W6P!{En{H%6pnru~E zKbvmy6eK=5lHpf01xF*ISBMu-u~zb z>KWg|FCDzL+(6?7E2+ZCxu~jX_bvmdNUr_X!Xp)NcnjlFG1Myt#oY0si?@CNnJWIY zz5AB+Wp=T>g(mZjOkXnOOduTb@jr+CT}+@ zAWw5jI8u8mqtmQ+9JM@xO*$1eud<71_@dh_6@i@0_fj20PvWbZ8^u_O!?#`(tQ`Z~ zpWd#3URX{3;b!hs591kO1sMjfoXXnefF~)1fo5oMYE<;JZmYMR!xFKlAW4d&B_&}; z`eClvmrc4$-K*LKiQzPGO%u0b2}uW>pT`Oh_{Ck3!3KDx98O5)(gHa$DnRK&{nGLz zov2^b-p3$5ccVn$KO5$(@>t?@B$^zKtjbpv_Yl&>=md}<>nW(J#{${h9`JF?WuA8B z5?bUsfK~4Yrx*;E+NOtImsWvW^_Lep_gRRr*BBRI?0bEJ55n^(`Ixkvq=e@_h zK7bk4sLaKYWbiu%(Xc2o^B=m(o3}B~mhMvf=w{vt(!2_neC%Ovn?DGpO05v;tG1_1 z?WF&GH+NX-DBd+*?@-f)k`5Jydrop)^U|;3RjoZGptYGxd9YSh?W*yZS_iCD*iST{ z-10kUVryqKGj=r8oI;5zG>QxNq)tmqXL$*>&Sy@33qUff-{1@#XDCTUbj`GGp__11 zw_w_2i|eyH4l0bNk4%*tzr~PpIah*9aWFe#`V{PhWEdkwnarQQjin-Vlw&E_hBXN6 zwgMq=j}}!>szFGwv5+ywU>EHd*xy@`e1>G5!?Vm5FSA z{Mm{mTDqs14U?Esrde2;Dxf@DL|=5NPCU|2S_SLg!<3rpp&(`gjih$;HroATbVfvu z4SdW`c}xx-g&#Wn8Hey=G~(bS)lG>Wvu@ZiY^~aKmBE7_RHMw{o80o8t|IQsm~jd0 z31wAQ&)9dt3>gyGm5S|&%g(4JySfCI99!}G_wZ+wGZ|Ds*aAa~)@EEhW%l;u=}rYV z5dmZkTzKW;1trJL)ZBojD#rahv8}Y7%oW_KSvs!DPE0?-6)0h}G$P#&!einqAv(Ky z;J18vvpHO@aE%n1rXP5#k@Z;3wiQuX%jB_KULD5bkvclsz0R@jOANOAfKBSzycrAe zxcINH-80PiHNI2~5PNgK6auww48|A(8NxEET+5B*SI<*F0TG77qcpzlL#ruMYh?QI zEU3}(&i3avsi9XHBVjd`^+RL+*a@Nd2sg+%wpkk}VgCG@&?Snuz7gSlQ+-0K-k%KF zarM4T^5|H~{$u;{^6=5CYuFWbpQrR3&G^3BgWJ{lM- zwh^XQNScv6Q+}U6kty$bX6Kz011pi5QV~8((0mXjj~vr4e$Ss1G^k$00?_$IndVpF zZuHjYJ-zeg{Q!!Pz8H&S$~U2gv{|LpDMWhRDyKCGtJD7ljCvda9P zbl+fIN#FxOlCipJh*27!%#fJHd)>3+gO@e6>c~QItn;TB1WUH z+uYb{LqAt=0HFMs77_u%V;Hs*onWFOsTw(5YRmIhE%Db2OxKyJvzYkb8Gy5a4ARRa zC1lhi7V7u>f;AHUEpLS9(?f9qm!t(6VcRBo-QUnTGz^EV z%1XJEMY@9*RaG^@Ojc%@!s8dQM$GM5UD737A z5{1zhU1iO_k?N>v#o!jtH?Wi{i%a5M&E2=E;x!9US^+P23)eNIDHBa9kUW#h_gDn+ z)lvw7cFnbF$s{KBI-feIg`u{Fn10D>jhDqgm+;%OrK^YO z8`w{f%EnK*k2H(}$)^fCp!;?p77{Gd8MrZ_rl_jYsU;~HW%3|Wy-w9f4F=5%picd{ zw1w?_mm9fOwosUeMKk7t2fxUYS10Zq74g9nWoq%<!MBNFxUc*y(Rd|Vw0c)t6|q?T33z~?RSCLN-*CH!Y}s70S{7HEjUe4~A9{)Ovae>6FEf_gaV_)y7H778l%n+dn{VPO!( z7(;_uZR-rsk4i>aj2DZIv-nB9TDQPYoLve($-BR(3sCAK?=GyO@30w86kkVZaeH1j zqvO-hKu!MjlbEFQ^?1WV5YyG6>FA*t`7-M^Ak7^}I$~VHtq@^4bbX8%Su^2ck>QjF zSit!RmR#WdN+wu-l*`j20D)7}(HlxmOMf`eW~oPl)HyV53Zqv$tgtV*K<9TrheNoW zJ-B`SnW?qU@XRD=f&nhGTbNKE1U;|L&@~~TQlB&S!?~a|?cRfGP-Nm~DX&qv2o>WG z8e}0PeT3kOHZ|~HRow1nOY@DU@vMzrxv%FaU9DoM@=T08e)47cLT#+xQ_o_e{$S=E ziPO}Av&+RcH(jOh^8{IzJ%LzcVZBLpU-~5S zFHjyz>=ta1zxZf=u{SWVZSG+53TB*APU z<`DHVNpPO^mR02={yf&Kz2=f$(^ms?YXe%lya7XRJSE}p>xK79g4QTokt~W+Z(-+f zx3S`6kdKnVzZ{AB%ovE!Mq|$4;r24%h?po0Ls+QS}tc{;^d0$>ARcgD@K_oF>0ZeTy#dNsflP!isF3RcG*zx zCnB+D=U859@lQP7!5TG0DM+N-41laRE_~PB@(4iVK^t58Trv>Gur&|bG(PQ&xyT;c;_@%WC8jq^Y>#+pg`_sa!NlZ?tV|!;pRW6~ zAR>2`u|TEmhHIZ=0Z>Q#7%LFKKgJ<9@s&?yold+ zy)+pi!$QUuX$}Xm5wD|C1KO7F&rfQ?59^c@HC|m9?(TjBIIwkk`L8GWJ=klWdCA^#xlzi=i!m7meUl}BuP-TIxdn$w;t=>v z#N`c3r3|L9f<4kYO0<2r?Vo4`Y$t-&@n=ok=#eDV@xNKMi@uh`e`#PZ2flZ&Ua&Tg znYORwN6;;p2bl?m>81RxT=to2+JYLf(z@iR+j#NO?!!G3?Bmuh>Zx}HRtJo8%RMyh z%m7a}aziOQior)g*sWRHo3=Q+*pi8$q7SsI4HO~MU4$};pjBWW+!@>haqjL(m>Rz) z_$EK9C?D^O*3DZX$>-OC=u<+u6a!8fQKFY6-4@gJN z5H>5JFRLwJPwPenV?y%`;g5-%(Cl&9M4aF>7O{aD0J7eeTlk>()|gPcu9Fdo{^Pk9 z7MP`U?K4AZJ^Nt4=ff%Vb+ zu3bp8mZ7F1guBuYR%$LT!IX_VMUN+el)Vf5V5e}q5|kZ*79mJ89|sdyK8q&r3Mh43 z5pk}AF4z`T2az2r^UZA%<$?&&rJH>6v&=|lbksqhmAak5PEut2!Y_uHZX6A1?4(r#2!r` zWX=W;Kk25Ll35tyxqf67WtW9#E?!CUv0bXudtE(b%KX4a5Z2HVog7;?-DdwbU@2=h z01vZOygv^Lu!u|hs9^0ex{=sduF&*i@GV*V_UPs%qFAn5n{;>b7~O5K;=bGU_@3V~ zeOnOD=rkwofLoU8fo0&luF&Rr+BOK~j=%nR#|=NMUENB3t)&x{%ln2utf10J6vCx! z)N~#bxp(Hdd;bMPz-kRecCs-QwyO0OdG4GAuB#d~OQL4)6))K{aKtKvv__Xi@ZBU^ zg}q0Fg3m^{=@k0@I-g_7UJuUcxJ*rb-eivR{(7xzik52;*IX7d2Ut0{-}jha22mbtuxrPz-H!yu$i z1LmhQgCNkz0qJ1p2~AD4_z8~fr5fiAduz=wT{WciU3Uyg<;~bR_YkDIf1FA6SY$C* zPAmn&sQ@n@4w~0xQamp*9;erTPDX%EEv6M$=#wYH>{;6Ns?lL~Z%xl+>Smj-S^M>M z1I0d1dF)y3l3YG`PdMMRhv@M}k&?A{8D9fnND)kc&Uc`oE`Q|6RvFu-!k$A026aH~)_I{X3lY z-)a1T@%}mf(9r+g{NFYH5y|~Skr4b}VZGRYr6!5jg6b0v4Nujy~3Y%ME~cGHU`w`TrOmHQ>uX4(}gW{~6(r7QLzp6aZvA zC{BOkx7*1T1^@(l4+H@GCH6lLjs4#o=3hxntWAs@?Z4St7&-hevGe`kK^?wZTR8q5 z8}ct7E&DdDUJMxk;Fkvg#QtM!M-vA}IwJ??b4{K2Rkk{>O+BF3ZB1nZRLR@IVZMR^ zyj5ncP+qIZYKO;sZA2)^p%~C)0`VlD_pA$WiILJdz74JB6e=?kXi)5TyB9~0CdPiV zbkbDn2HN<#H|>^)4|6@zPfJ;^Er^@YY_F>EcvcU0O7DlaLe{R@sxv;%hx3=_$yPTe zPX72yKBq>>iqAD(CJUo%nHhQN>^L1Q^C3k;m7q_Au2rV`(rYNbOIcp+AL975DA_|) zKJJ{oKJ=dtALn|=@w*!5Qt9s2-8zS4&z9vQbHf{^WY@Bl<;LELI@+{m%QuDwRi?}H zZ^)N=n3c(gw=fP}GIu*f+=LtH#DqbdPl9(PjRH#&TWbd)LECsdpxPK zX2Up{9`Q6|{rhfdx&<#gEbSSY^bJ{hncpT3a59fa_j%mBi_4niTvpO$Sh>e;v`%rJ z8`uvR72C$9AJaoSPac zTaA$Mlek_7sTEo=AdqzS(x81^ZE-K#Si&A{9VhEa6I10HwQf~ybGAvn&d(rMDW#^; zVFxQJ=&W+Gxd}K~Of752kQZ|^8hIMgERII|a2pG7DqrChJTEh!&7{)rXjxIwIou$P znJ5+F$F9C>RxatZajvqNE63S)0Fl4q=T%1F4$vTgb3DJA0dKOM1HZfpGnm~tS~Cv) zh1jkLvDo!fy41Fr`>bG(cOi{mU}&4|k!^B*VJpB^M%KuN+3VBTGV>5>u4S>y(@)Jl z-wI4P<}KwshiUh`DFWPtcXOJificzUJPjP3)L{oFb*bIIAz4WqEGU+4=xEA<5PFn0 z%Y2_&jlI*Bp{1Z$Ha&KUzr^X(#)7A)l+{qIHH`+;jiyF`=fu}05nieOqbVw4gTqsS z`&B|uVml?lODcjfQ*Ik0xcMpmMHGJloQ-}CPjgew00s&`f8(A^JcXuV(~k2JKZvr@ zh@n#$!&|R4Q}xwq)LVrDg7z{pD0Imdav66LmNa&#;5gItJ&6=`X|lB#!TJHWi4NR% zm+&r}@)>3l_O)k%k*aDH0*MKHZH8+GIRKiuiBw@Y9-Jo#{CP-_2QY#Cu)U;qBnbED z4$Lv*DwC1s3(wNTwrNxb3Qpc_@Oj@RQ`qBQIwT$%oeeg+B=}sB+;^SjcHhvxLiE6pz z<9WkQhEuSQRBP|9RvuYCLR%U?_#)-OEBCIU1B~SK@_iwCx=FU>n=cGrtlaA~ll0Ye zbZo=7(U0tl2-Wk(cDWxjzZ}r(pLg~Kxl?NVnjD=8=7EEA?xV|-uk_VUD?_QWDk3(B z^yhcB_n{({)tosQ=!iFgr zc0zD_QN7ZqtiR)LH4V%C@`#8Ke1b1r5)gasc}s9o>Fs80U~ST8C*!f*EnxFFqrk+m<JSAZOj^vUTB41e_Y$b^%QX(MqCnR1Bc7RVBYO)qy2T4dd0py^Uyn5hA+ zaPY4iQzwA)!-;O>5?(uH^Z<#6j!STaoT%yJgEYu{3IWhp`@y0dB*J@;ksuV7rNRRxZCSiw(lv0XM9J8a`SVds{MqBsqaWx{ zLor$VmJC(u<6c@&~Imb5r4LATU`RSW^ovHMY*r6FgFQf*ToH;t7gETZO< zi)oc0#HTU;g_zN_+u0hw=U;1|2|fp0m&Xd>akL`;G~A}Q$aFKW$F zb6T?31Y^f^kPG9e8N?$$N7Yt5tZ8o%H~lG5Jn!Hco)WukOBK-sz`HM`mP-Gv3LkQjnfgm4=vI%}H$C7~ z2jGoru_*Uj-t*tJhn-G#NhNl5yy3uH+MXqNBlQj3;UnU*mj;uM8g(DPH1j zR6M<}vX@`_2lE7!tDsF&v?5jKxq6ALNEER3@hK&QzbPv_j1=P5Cx5T{e7;y8)PqPP z&oCH&j8=#0U@lmSdLO;ah#fA=%wTzEV83n(sRnISYzbb2)bpbys!%U=e{q7kHD4kA ziN}<#d!IT4G?YWcR&`Q4N=40aa3&EG6AL8L!;JY3l}w>T8fzu)+2c$bdU09%twUKx(h}_ErIWK5D#9~ z2pr9usI^NbBakNP!|74$-4LlLA5&B;ExYG#3a13f<~E_tqAZeRjbL#hCtbPIM(hO0 zOjJc$A1qr)#x+!OGTLv`8n{62>#Z(j_%}-%Z}+36!?=5THKmMaq0`J;d1f%`I@Z2> zYBx-u32ugFP4}x(#1w9;s5whLcFa$U#!juSsask!v_-@%p4L@ z;$@aRaqv%k81Ov=7%+w(~oWEzrxH6nBAqX+6un+J9b;itQ z$;Sq5m@C_*`pV)5fJ@MJMIg+8fE#fdmr%e4_A)KB5gB?4c6VYP63zV;ItLBoJV9Ps zNUbTVBkb?EwjC4Xct}_(LZp=`@M?YsLk~v($V+xx-tXvLn=w&*t8UO6?Ue4)kg6hj z&qrE~4`diJQb0Vg4RqN-sCH5Y9W^Nxhn3fq2iGG@jF#)5XTMYuVPqz^RM$F(`S|#F z;Y2hm92+9htAZgS%A~r7T+u+0nij~$Hg!l~5toF3a2ikI{+=TZxGcd%b6*>eUAl1X zr(djwyc*3xevnG=D}j^)ryWq}tb0{>mKtgh=I8-UMcK;VggHq8pLJkv9`&7u5OTFq znA&6W7h{2%!?LD*7HaynGiN|aRq42_giQFzm^Znt+B zm`48M3^tPdOVDFb_U2L)Zd@D3(L#+r_{z9Dw4?(g@9yK|=7u*~LP4g;z2h0?MlZ2{ z2NAf71W83pvqR3Zc><&c0=8f#N+6P5R!pm*&A64Lq116b>Ej`uaI?C2%`!qNSDfIB zNX(u0WM{MdKFhUfP5wZ!LrPH;YXjV0MLES@1EEl`asdG=c;Bwh;5poUYP|*sso-Z`2LnCK4?(@eMzk&#Mu&x@s(ers{4ZWa}4eyyOzPptyj^Qxe zDs(OPrJtErkq{N;426|cW>KOiLkdxqwc)Bx=~p~=K%G;9<-At|&9s3BWu+zz)Uj6P z?C+dan?tTfnL~}P)ExTyHChC-R4dq;#v9sb1391#Vzbmj&T>3)XE#79|A5TgWQel0 z6*%Ss6AS;|x}bl^FfmsR<<+$otoN{#@p@yet4$3bmQgboD1Mc$m@SIC#L@mJEPjeJ zufK*@Len0q=<6K2Bt&r0*(y0y0VqxXsG}mDJ&Z<3W-)YpkawsfmrF1Lc3}&}c(VH4 zEDiTt=R_T#PwSOi96{(FS&=>ig1Kz=KsM*o?ixT=ap^6Oa%@~n>p8_ogv#}!5c4Lg zy~9p=&7hEYvZh##6F1~$;pJePjf`1pm{F#!CYoGG+*Q2V@tEYOcEpL(I;!sjzXSC8 zK9unUA-Ily7IMX0KiumIzAvwWoDM?{MRJy~DXX-*nuy48jgLhR1mkqWNE~ux>sP3} z!LY659azA;VS&dvXY`MA4Y)Az1^;Xopxf@2Fwbq3UF`;AW!y%`77=d@~s1 z!7JIFyRTB`aUYxZ2&MH#=%iDVg>WlYQ}*^l7cJL$ZD)tF zBtS9+d~GxJ&+zUGHlWGiNcQt!M}+3wWDts{hz-S4yCKx@1_o&fhxLtrZ8p<(+}hO1 zd&-wP6d5t=#&O)>wjFGZiq0U}UUFYbgnXauQWKzx@s{FHY$@gX%6XSre_0{=&cIB&=8fwWw z0Xo-0*iR&Js`CAk{5i#=m|5<;Gg-AynF6HPKED1?5u(l?PBE2iHT+CKos^`VS3x(3 z-%1?`+0DoI0Lf-yl9C*FG|=h~%k^fT&uoJ?*50ut`^(4H9pj7}2bkX+BAPh|R=@1p zL|wBDx~3cbe|mlG^T;;X4fMCv+8>#!T@VUoLI?n$A07Y@{f{$M#wO0cC#v?PY(wq+sy1nFMd70Nk{X7MD zu3$xD``a=c!mxSAoAc!wQukX(M(w$A#(fI7NxG|yXn*^ zHg)^5*4Fsx$vw?0Ye2`xZ9+qu-|m$`v_I$}l{Ge?Sf^(O#gAUWwe{G4+j68BE6u%Eu8<^Aw@mF=(L+v6RXyUD| zGrw>c39y6=clzX$?#d)eEbg-^j@0NC!Yz<_p*F^?ik!#`Sg9X+j-m=W{X}h&%yel` zD=+qWuUEY_hjDfvccSDvW50&)h7@BR=(F~*Oe6%xM|o)a7lP6w*2YDBT(`)lwz6N% zLcI_rX3MG5dxi9e4yUx+Er<5hFypZe5mHqJK3TgPfY-9eACO*Iz6n|~((N!vf>I+% zz}nNh&GP0@wdCw*PaR=i^St>(JvsY!Z=69Q1zs&fxi%7&p*+Ms*3_;9;NmjPWPg8S z#D14D@8HZ}{3%YOhrqr=tmM$>TL439zqcs-T{G1L2(&kane!(QY)nGDhFu>~!AA#I zc1}e-kSOT)ts{cUJ5=XDks5l#8dG@XXtQ6VnWV@0)R3U1Tq$#;qzvLHrsG zo*oDcWQWBYJVO3(^j%NKudv&+)$@YOjkx63H-BYHAo@EjD+c@6FK?EJc!B3{v5HHdC^|NVvp z9Ynu%)Edklkd+@-ft!(h?n5`VW4f1W9hl!CESs51+}*0O6{fBe^8<6^NS(1l(M9V} z@XXWZ}N*r>4W#oLw4+`UOa3P`T{ArsrieMh%c!J*F;JS z`GmKC$|SY9zHX=I=j>#)kLTz1ZaPBt#-~TTHa~g#p@yPy=ulXldu?li@ddKY?ME`v z>v5U%TN}}+WQi0L!Qr8SPJ;0IaFdEfs#)Dk-Oyal6h~N*LaAJvZgbnB25G1g)lG-A zs;uQuS;M!NGAVC(RD*$T(P@nL0u%-ANavC%=9};=!dt&KUAD4g&1F;-mAZ>5fnc`r zU~7fpgL7uO=fJM5oba?+@UscJomUP`=zBqYfqVcH4?fsSuPea1g!isDHHZ?{Gs6Qj zWdF@gPo+GlkTv9Tm(bl2Ax6j9Oljx6ql}jd)jd9Oj0kg!D=UT5dcB6Ay+nf%q$1NV zitx2ZC1Ix13zhoY5<^2IjYxuR-zj-5!j8-ld^`_OCk>mm5fe9}0rYzC@u;@H6XTTP z*slLkp1|<&#YxHT6HZPlPsl@nShfrJNT4W=5;cqKSbda;`D& z_WQ}iID;|El?L#8_~u|h52+k^IXNssJdZtNF^Wswd&{a(CuxmYQO(n(k6aK{HHu;) zWk`%3O0N5TZP{jnng|EMq^nfJe){SLPehtv)7;@xtn3C`3#=1Kc*FTjZjsWWWNCVq z(mJc^rmBUSa`Fi5oaAj$MOl}33k{gVV$5V(S#FmF>Vs8koRUe=b$D5Y)l8ZEMqa5_ zP{;Q~3@&2m9sGz^S6(hXtR3yZYCOs{GGCvDg&bmub;6sS=SvLx?GLruh{c`2_ILp^ zq{}^w==4z-yWutk$)CUg{(x6*i%OO91HP6iqs{2fR5-0Ucy_h`CU4zCVMr2P4u2f> zo%Q(W?J~0BWq>El9x(inj)G+)d4EpfH1x`jE`LAwjAcHK_{nnwdHeR6gRD<<;$C$U zASM?KBdh~ndkF9|+Bs6ioZsuE*4s-IY_0owwI55n_4!nM;)Z>pBfPYY+TF6ONm&!) zDMdodQ^3(}6@$tujYS+}n)Kczw)k1WQne_sUo}-RUs+ZIM;+;9S0f-hVo*MmtMerp zEE*-V*ePArq;M3R-#iSf(T#P83q+JT)jEgR^VXS2#LgfIAgcu zC&EAP!7vxf4>Je=z&RlRApRfk!H)I@7Pi0d!jYO<^{Z@g-kW-Uyor^4jSWd{3dtQU zXJMJHrk=}v%vI|c7!v--o)Z2*L&VhG-dz{qBw2llsXx9*6HS@l?Vnbs*XAGdEpwIY z6x(FqzZXx;-$laiz&ExO~pN*)fJZu@xIR!$k2%+|cs*Zh#} z`Z3}bzkGH45??Czs$Z9`{+y6mBqtn=R}TJ`>DBCILyf}Z$%n(w|3i(Ri|n1K_-!zp zXY=EezW)~C8`JC#`=z^95Y3cjz8Bkl;wko_xK0xpHQN7%Ucsgl^<8In{F_%&{X{OouBdV(Nk;~T2vd)j1nU0kW z(3!ScyBZ1%#DTjx^~d7-?w*X$kD4Wi;~}h?=kv;qWAnqZx|x~X)%T>z>p|#XXE)$l z9n{jR_=)T>n~RnL<}Avqx~R(*xqK_vA})}}V=|pz6EEhcul)pFo*e4NhgLn<%LY== z_2_8AQR9g*PI`y6Rf_5Df-KKDn#fAwpAk6be|eV$)>`*I%-$bTMp%HMlDw8$V*Bqv zY~ab;J7acYQO_(mVqNmCHkl;Kz4U}gwS2Q`t9ja#+JT-S31?_rp*~qL>>?==cHVN zwv3YdOLGS4>3iN`Fi`-UU2uGNkYQE+<}>DH(XK@mNXExQax}e<_v=M(xr>j_Yp>43 zlZW%8PVKTu*%Dp$+IT>8opH05tyJwZwVn z=F!L1GAhST9n~Mba@Db9R>~$$jSHfMksD{Xbv!9H^)usbl#l7AKcW*0Lc}0-gQym# zBK;KykCC%Xq$yg%u;S8b;VTxN>sHPlrRyuh8YxeO>(|fWj7_VNex~K2gR{SzxezK7 zUVK2m350akaU2)p52Z6S8ALmyB3Pf+D181f&n>Z~*%t_pXp7;CWDiLPss*>nc6ni6 zq|dn$>J&sgg*ZGnM;v$0Ai(FeUw{HU!l+{$AiImZ&`4oPY*gY^4U0NhWlu-*K{5F-+_Y>!Y2CV8zsoVVY70t|;%Bwhbv@r=GNnqlDJPBf)fg$33s`Ga+4OmDaE~8XI_a7FR zfR%1>=i@OokYG#$mu6Z=-Gld*)?1K+yXO0vUghz+^_vj zYb-zN%F0w9;<;znDO&RthJ#dS;oTI|=S^if&CA;6+!{B~P3$NhL~AL(F0aFQbQI1% zO1+ox<_O{>Z8+2LyH$ybgjMV45yJAoiUVQDtSUX-4WSS21J^2SIlwvQB}`d1 zWNtS=lRGm2qYH_x$_GaSSacVaSCFtr_T3%iZfY+)Ym(*y3ZB0q9k#V+w?RbzyK-Wzb|$#Ul=B9|6AVNAM&Sd`0@6oowW zlmHWMEt^SLu3%Mjz-qu*_${Eag(XNwv0i{r1E9vPjOWdF=Mn zoycrEI2M}#6g)&|oVYkQDY>mpZB~HW<4}-!CiZ2>gdq=~l!Hc8ePAu0EwYBTi(U_Z zdX}Y0C@Nnn5|s~%fLH_tO>m~n#rR64f%pP3#;yxRVG833TN<%Ig#Ecpt#s?#9d!kZ zQU;y@%gvL?6d}pGe=Mt#R+aeoJAr4Uxeyh%$g&T8W9hjhpCA&ZQyL6PRvq~uvY*yf za?=&%mN#uXB|rg0C^tQz?HDs+N6040*Y;DLj1)8s7&P)PqV40(^}JfL!&ty6y{b^q zTi0hCOudhBqvY&u#m0s-ftc=nJrgv3`pF^4EQD@?c;X|#0zq<>KtDhdFRnS$s8wpz z@OKjsV!W&Y&iTFq!lf1b0h=Ts@$wxa(9FbcD_Fi1r1kJQTI8!~{Ij;s0x-^fkiQ)+ zvJ+s3GW*JcS{rWxVPU2m__0N*FyIpUc@c0hhm=7A-@=U3kToan;eF8#aP+0#c}JA2 zs>UzOLD!^$ZOC_LE8-Eh#0@q3uDNbcg$Go@!wm6uvXAd38t5}l+q_pTAWVa4I_DQ& z0v{Jyy-_lg-vy#$yG@+LOKgSNRt*EW+hGiA`}EV#vILsC+UhhgO3dG1S`xx=&%mFc z4S|;jTL>cT@yYla-cCD@1=o*P%pPmyX9U68kcn`ca;QK;PDx)G0~bVngnISr1k%|^ zD(yOistxvV=C>os2Dwv;#%(3IR6TCLedtK=>7F)X@wriYY|_~rU1$jJ--q)Lbk1x+{*}QKr2I1}8U?WMe+`7^8IcoMydXUL2QrhoUp%D+{r$H@s z;vy;LbiDM~3Jo5ROtX+dh}l2KTwL%!w4E#Be+B1l;-{Wg@Xl)SNRvIib2ea=06a90 z5e{N+mo_}XlR$dCGYe7}reu3^h@;+*nzLFs1@gNJL`3bCW*`*;;w-FNj)j}_h0XV5 zS4--gd_fUUh|AV46AIIab?C2yJeX^AEc;L~c!Z`ja*8R!Ei-t*_$YhEFPpDII*X_F z9QzM``oW!%?a-E@R;BQ?@Q`vQ6c<^?jb@U4@+f%$d3EZ=0Zf;n%`5L;i6Meqg9Ec1h2qe?LC?>lCJ#>Y%PO@R2YlVC9$d;3{L2y?s;39@e zU#+i;lTlCBqD`?8$IVFPse8;_mIhUf5~svkXLIaEYdRv=fzV>yy^6j%30kctOIW#nZN;cF>vRje`lmb>dzXXW{P4J+Pry?(C}<6?PMt z!^6xwO>@$Qnk7WTz0b*mv?ScKwb;9Qq)dN1J>eIt(|p z_7p)RAQqR6G=nAjmdF7`AE#@CN{4<{i;?c1H}~F5ESs#r8zTC1&ul&(KSW>PJzn4p zlRgO;!qr%d6ovj(>%GB?qhC)th$jrZo9I1S8a5<1~RJ)k=W8ifMH3d|$ zYQR#PTPd3nuwr6eO)0Som%oGMCKQ~p2;Ga`imVoA-s;QwT_>vZk^qY^@`|v~d|^d5 zOS}9;CRhfCM^w&!J+b2>>WX#4>6jA4xXyFO9INm8gm<1p~Y`dT*th1=ZmUL8v zgxJ6V!04?AFo2BbUV~;@w_UI$dCFEi8?N-R1#}1+x2Inxx}0mg?@TC4DW+ zOM3Oi?!tx|#5O5{gap-LEbOySlf{aVHs^uA1W!TaST7&3^wpd`Lv30N`Vy5D7<8) zmU)-2$Q|NTB^xLTq1>BMNFK(~&M8V8?Bi;aK~i2 z>zyG@51$&R2>Cq?*H{H9FSm=*{(ke<>yBaM+_?uX<7v86Ey2h{V0EWRFEZN@>PQTR zG;@}$3~9=PqQz=MY5;RN zYA9SOG!0Fo3WYc?Casz_?9WB!Y4--%)sIfK*3Rvgof{R4<6`(s$_F?!%eG&ib-yLG zzn5uNQdQozX&h?VxFS+-gfL~BlrMkzdRj!2+-}myx0ha4(rmh_a^JR8F_O2z&eiUb z_#J`^e_KzrjlNU{V$oMu-b6AU0&SF=UyxZrSLR|)M1L2ayxuE#P zLK9WAXd81_k`2L((Q);_$*tGRFm z;4YnYHRJt0zE_)n8Go8hH~z)dCEcQS?Bgv#za3(iD8UQf;12?D#)K}gMZ5%{AP%NbHU%s ze^Ltnx#RCz;XmwmN6O!-|KRu!8s7hn{CDl}AND(j&TrL!jq%Sq;NKDPK>kcijQJ0U z+5aZue=Cc@{!kYGKdRzCpN8Lc#edlEA%Ft?dm;QwmGPgG{H`NN@Qz8F-9sKJPq`!0O{kBs7optapCh@QF{q;QQe^UO>gxTLIcX9qx a%KuKEDM*8X{|*fG+hqd+0ASPn-uf@dbVL9E literal 0 HcmV?d00001 diff --git a/datasets/subjqa/dummy/grocery/1.1.0/dummy_data.zip b/datasets/subjqa/dummy/grocery/1.1.0/dummy_data.zip new file mode 100644 index 0000000000000000000000000000000000000000..77e2a73d1239d18985ea00863208f6838b956f3c GIT binary patch literal 7371 zcmbuE1yEc|+J=YV4#6Q<2yVgMf_s8baCe_!umA%Df?IG21Rop{EV#n}K>`eJ2`<4Q zK=_j@yLb2ByR~)ick0aasXG0>{hhA*s=J@3s>mqBfS-})T&VR+_MLY(8c&=vhYq)*byiIKso^c@Z@jlf~-A3T$Y|*`+A+R^MVAe7wl+$ zc3wz*@mBD}aoL1|SsN%5Qie0B2q%lXRULZ*oAPW~zzbWJ+1*F|CRiZ+dRC9ijggl0 z1aQZJ345Ls^LxU+1&H2J+-d8jqqI<8j6KQ40ri0Z&E3QJCS2^`m9p1`(%N-sf%i!{}s0PwSndQTg;X5eN1;G zp|LrMV6YWzgmiBuH+Ex8c)pX_eq~Ns+i~cAOUn8^7YK&usToz#cAUN)cxde2$}L7) zLrDKtPZDd_`^B!z?1I)vGZNs7S`Zzwap7V2Fe=Fd5CSPc)#t6eQfReee)$Qu_u1;? z;{39de@pB4%l>fF9-%n0)LCA#_zFRNvQ5?-e5@=<>>F29jdoJuCX)az_~%ED)sb+z z;Ly)|tav-%@omLtmU`3n?^QB6)FYYb<%QS=GaTwvbY_Vy-Y*aXp^2~6;ktP+@d^*o zd1Bqx=kT*f`<-|d!B{d~(RiVFF<=f6?5z7A425z_k`ydB6R=84b@dnv)h5+826?Ms zQ%q!J96e`o5UEZ-`+Cv<#@n$@F`#nDXd;LhMfD*t{~MBp&a==t%BmOA+KQNrmwK*{ zJnBx|zUVQ06++4gPokinhfxyN;X`9ybJdh}e3r^+iG;zAgb7pYD$HxR*x5`hkx?AY zZM#WyM8)mBAF5JziTUMr?#hX8St!Zna8HJJSWT_LQ~kH%5>_qU5cEemAY`xNrjW5r zGNckAL(%BZxl{Oj2f@6X)vGn5)7k#gcs#s%&4*2dOtc(ohmUV!D#GEfI#9~<;zF8w-Oyj-h zp3>V?m%XXsbfDebq)w6U2y^Zncq}{^LvwL-8Vnop^V|)!8wFzg40zEh757^r-|NS9Th8J~g^~C~ ztK<`KWzLB&6DQa!YF041Z`^|E?)bu4AG%kkdPM7`{qKeqw2_y;Qa#3tNt0?GR~L$# z`nuoIei^V<43uGz8X4M5zY{-sWBdJNI56kxZa)1E|Mvp_Yu$8R2FrXw2LQtH0RWl5 zRX44yy?&O>Z2kGzg+{^`ANAdC*B<3EH`f<1TYAl;c2@RF?-@(svDAE0kIQh1`t%4r*^F5j-$B=$*VrP^U^M11ab@=eUNE?Wujt)!N$ zYNIL}kv3WL(L6bmaKET?)L`FEX}`&`9+_VrGW~u-H&aJNki1X=gtubxTSn*zOEwfd z&z{PrQ7+}>o_l)`4%_=yV{bSlny#HkRncEb1NR{Rkoc`-sJV@Nj`#4}^EQZ|dCRK} z6S$F$P`ZB^lcq;G`{Ebb;%2WxN47b^kJsg}KHc%;rs?C}P27kdrK*ka@N(VcrL|sC zWA=jxZYRyHS%jkH%wP+Gh5L0i7l3?96Fm<%(JY)U_!!!=QZFWuMA{hvnCGiVRUUZ@ zp;t5TN@)>pls|>bkW*b;0&rY=DyG&nyulq0yC`fMdUg9ls?Nz=zXmGSEi;z5#>RFb z>BI2wbAQ&>IksY00UUoIarhT9Tq9+Vuv6;J=?&M~`)No8~qoryL=+=^bt+d?Umr z=Olz)!~7tiIrJ2nH-pKWkJhEx`no5#4yu$;7?bqBo5b%*l`yq;kTFZPSWr;mAS%L9%VxS+0u78XeIR zReaPJarO?SwyG~3BPY$g+jV->o<}rnai1w*)q57P!esD{@CbCQaAVGV@=D3=&>smb z!LDvgDvqzIK+Db1BKJ@pgfg8h?_zt*FU6`b|9EKOu>)DiYrrSx)we#kFX->^M~BD)Ip-82Yn|@SAF?Xd?L;0Q1DWK_jjzwN4M`5Avcb5l z$5h)5{OTTX2PJ8z7x5~#vKnV&AY2{TXj!veSxT6NQy5M#GrZCcP1nn3UJymZ0gbK3 zEAjD|aQ{tD1Yb8xd)46|tVhrz==c zR4t+eO_RX5$98B_Or}>=w)ZgqfM-JwivTsF*Y)X3X|4^XfO(dX)1}r>)5bvy=5@+K zxIC+UU*NiTYGeSXbM#u%ah6TWXIBG;ogR3)(e@`y^dRA z8FX(c6_3D2NULiEonPKthWkwjwIlPiA#tA*7?@~yyY&M3(F&U)uNkund?^%L;tA-Z zqz33+S3it9M4Uh;YX*S*&n%=F&vKN62)FyHW!i10&!sWv+=kDUv1P{16l4c4frIf+ zBqj3o^K!|{@?E)L`#7wY7xZFk+XfM(My8A^YR@W09$6yOSEX3K#dB#Lh6<3d#=gku zRC?>0#6n}w@|a7y>Yf!}GYCXy;dT^+q-TaUd9-n8;tYv+uZcEA`>~K`W7%ru3Zmn! zovg(V4#8??qpRH7+0VQ_UUc1zM00rraq!l0Y^0-jS5`$`F$c|%JQQQ=tK{tW*A)b6 zH;fpRwIl-5FG_ZsouXie;c8Xi$_I}soxinlMH$RoGlYNiGKH_f%FCLmBj#$i*oPts zRVF?gXI9Ns9_W~E9`}+3Zy`@zkh9{gg#-`F1_?RratlCp7GHg0u_%^U>_u6nJ38)` z5vSbGyWAv%vOjx+kh3i#N7@#z^*8KzFgD}r9xBF9Nfg4MnI580cw_)raEEgyC_hWev?GG-CzBplBZ z%^~SK5BCerlN2N|Jk0*=M)`s$PMXze?okU3-mOsH`p8bec#_r9O?cnQvK!hp>`5dE zglyD9Tl|{+%cf6++VJwy#v4gvn+*);-Xdgho$(^07p3&u6l`zD*9q2Et~0;WMzJ%6 zATDbe>3=c}9KVmNUU}^D={c@?_I9NV`w&Q13qqNW1b%nv?=YH4nK@siOOm1bY`5l+ zMW&MPNfKL!&bgZv94W13#5!d6; z%~4i+4PrAMjU7x|cYN%}X=bvI_!!@ynhN!9lPbKn5!uR1%29SMQUqy4nm7R^pHQZ> z7*^pc>c{nk`P3kPN@4t-RvBgCGd%#ppRGC|pvFF>mLg&~P62=80cz@vyHAw(9*N?) zb~-T_P@YI!09BdWy64J!MuT1;YWFl`V5F)M!-nFNhIW+2Zabu~os=^s zQOwb=(&32`$%GnN0j>ax11-h47Aiz|3q`&;V_S0Zh=BxEqJZ2%R(V34+}8@%2<@t*?;)@6T(NSOj^8FetkOI& zN@%?Ox4jqG7dLwX69Awm0sv(H)*1rjVQ%m8a}|+npzb*@e82Uus^cx8}|IWJmj zTL$ON{^~;MI}7%a%Cw}P?>qh@#o}E$A%>To&ENfxM53osKXOWFu{$0~ zl5euste*Hp44cui`tVX!CEg$#muqQnpVgPt!3MRm{pWUNW^=}T4JFd{TNLrUYFf%fdAVW|te4xq*!ClQlw-r`^77>(;xb>rJRy2RoB>=)qcoES0{5~D@M@TIKe~!Eb9xb>Uh|!6ab!# zpnW@XC}N&Xnml+?^iX?qm!oW^Y+uya8GCNURhv;i^UX9>A(?S`y8^;kZA)$eOTF&t zork4Uv;pS~+-qr@vH|zv8I9@vPz;G#q_jh#_Y8Z=_x+8s(!{0N>q%(5 zJ&(S}AV2mm`UVebyP1Lp#4kIz-X6Eq|4uXIhfl94Lsf-hs-` z9n@*??QvMEXwV{x-}{DJS>EDABKAUU^u{2$sQFy-#oC4i9W$)Ogi|sOv`C(s?g3rI zhlrSSJ2mk%sb~1DnfJY;>p25dgc9SMZ1GKKH*-i-#av&8iQY^2P_^I}Qysb}10Y-&&-x1-ToY_zOKtSaiPp$dkIU|!?39L?=684Lr#`Efh% z=Sx$o6%xvVWXGubOWG}s)WRhIJoWqFFAl17i+zqb6hU>z{PMAiv-}VG7$V)~J1_Kt zqGFW&4DK=c(&yIf#=0G%y$!ilcr}h3qK}Hpqppd^OFL8$KMv;7CGWV;OrWp@OC?mG zB~SD`8L;~NOgI~2|1P3YWc6j$#hmY6p!j0uIQB;Qo2dRMP=Q`P>?WQ%a})ovGbD&X z*YsQW>8lS!57X?JnVz4^F)c;=A+Z3ch`nk2DD0`dO^?KNcFy@dV8!L~@L+Ml;Nks^ z1i$O(^jX#Sh?8PJ)bTHIHGGtZqGv;AU$M%A79&XHm3jEJNCjeP7Ux~$QFSl0p6GaG3X88|)1l_I)sdOe zOawthv)owaNXXKW| z{K+;VZAo#z3CRW-=d><|uu>+lOF}<{)N+=2OEyKU`E~^*2gR2JEC^J9LEJV`7d4<#I#TenL!%EkUsTJ?8Y0xkeGryZ!IPG%9oHLRV>-^)*<(^&)X9b zT-s>Kto9HS<_x^UQ0x8_CvQ1zPGaLIP5ZSp--Q!U4nFM#hhgR-Xi0xFq*exzX zQ%;OI&(gtK*nQ`cBPnglOZREaq=pQft!($nW`-6llM?-HtqP6+P{zKf?B4Z`!c1n@ z=ch_c#+m9(%{o%4DS;Zb*tqYWSE|+64~d()oN$7)>%s<|O*V_hDUv<7v>qsTQI-ovakBC0s4#aT@tshiv-H)2zxvu@Meqgx4CWYlM6)_8VC z1#c1(+5!@m7H?0YaUP62WMy`aVwySAZ0u%%a*|(H=!up!=uW0NtKd4|Ek4)V6$fr! zG36I=?0U?m`4-66^|YB*k)Ec@6FwBM>Q}n3lGVbQDq5kE zRAUjwARIRXuH%2FL7(g@+0bu{@{pEdrKr2JH*LH;uuL;eTFH~*vJKiCZFuWaW3ahX5=em}{~ukkYllpiSPe@Ws0 zahg9j_(^L1X`uVp4F17we)sv$ShARznVBWGn3BJ=2-5=qg8zxEld+={ouQ-4xu$Hw8XH>Yk$R3-sy$|9q6y2a89m)l z^tmLN9GeT@{XRIHf=~!44gg>D<@BKwx^$++rM&6Bp|KEu5Dh91ulj4ZjgwB@xv44q zp7wcL-YH-6<|Prq%We!PB#>t~dw=B!NR+X0Q+Vw7ojx#X(y4`hb=endH4jyKC-?eQtE#BZ#XB)Vr^deGr5Oz3BH@VrARU7^Wu+Eg~P%b4LMPKCZ>P zlb&#&^_&QD7#O@1_L09fH=W{nZ(Gn@BC|pIcQ4VQ1bN$qA8jyy$dqTk+q}TUe-aCA zP~P(2YXWg?e7r-YKUY%G&W0SXTw{_bI74n#la2T&4R;sW;wDx_~Hr0RNy24z!PszGY2}g(A#wlV!)>fftRkb(h8-A zVaTmAyn&gQ#L6=Ul-HmbglHrNqPA>jR$5L6{A{8#SAxZIaErH0#Q>T@tHgq#gxDeH zkR0?hdYp2(Uy&xp60iEfe>`-cw)b{-xAh~IF~`L5p>ck5tO|Nh zO(rAt!V{1B7>Rm1pm=LSiEqaihagi=nn(UszalSW(uBD!SS8>qq-oeP#v>v$u}Df= z7#grXQBlyM2HIDn(JC2Jdw#%m`gBXMrZnUS8keMgapwh0t{7HcW1hD_ z*d-1>U>y5unnh$GqnG;UJQf)&jBP&{@3xVy1E%lUPm!*YsGo1uY#jqn!oHE3d?5Q>SA_ z)&>^o^q1!4_8rfj1wK?N^P7`%U=vf&p8L+V_?nUK$$78Wmc`oj4Ci2Q=i#Rg66$@5 zM^-!8g{uF-8vaALLm}DgAph*8vihfBTPbVA)Qs^56=4L|9w8@%Z4x;itO7NuQU z(jpN)O+g3-sEcnN`9vA17#cqg9kt{yBbB$g%F%P;t>!C(z+eyAgic0e-Zqf2QMMvP z2}OeW30R#WJ(8f-(7Q`AM^a`hFu)X=+jC@cX!5R-FeVGS=2_n^C6K@8hiFba+vUXH z3)E3w-5sr0?m!A&jcY23<>}G`c2oci8u(cQ4Kz2(v@#ORg*Yb5CHHr>qZ);Gp!5WaRz__fGj0 zX9yDQ{ptF=rliF2dEKRAaeu}wEJ%=`gt=me7uT2K;v1nO9CcufM~;o#yHZD7yAX6( zp~U?}K|HBQ8s+*T#;SiuE^J#BwaO_S+}_}uxn6Nb#tOj{R?Guj*~H^viyH@Bxo((J z2sI86otIA3A8~!1mm@-0O>}00ad2p#|LPQMx`>_zUWdaVV7{?!N^{0Y?6maPi10OU z_SeFz0QG!yHhO+p%J|OwyR>k^1>2?@fzB=2#2FO-+UrGE=-Nwh4JxAq)qUDPOw%Ft zx=&&%A8)^M4jo(aG*47OGxJqG`&1;-`lZWmy%a+F=EMqmtUi}CDwexz6EJJ%r224Ji^cF|3={!CQsk@KDs`%AZ zV9L4;0cM>!g2ey^86Rmwyjo}F@+?yaBRfpp(c&0uKCglcQi2(;d8~$xv6Oy2-mN4dks{QK!x)9h1FUWC90 zvtl8$N)_!?KiDF*=GNm4?jUf)EAi;Ah}vpJf=G1C=*+Wbw(W`wFGFIS?m#zVAI4~4 zn~mQW9fZln_2Y%mm~Nk&xlkg$&bOT-WZNvv_YvbfH<#9y&Qxp*d(U={UowL}O9R>) z{Wnh6JefT56jO5}|3A%l7p^CcBo7z&#tiHei`F=>eFL^+#ZC2_yYut&SBqLrxtpN~ zveH;?NB?Nz`@JKrt>s4D>az0Ds%m!j(k0 zqk9u(_O{uB!^QRek^G6JD=MGV36F8_EN%rvZRv9f`bcBOY}%WO28}A)WRrT0h4QuZ z0aHQ{Olw)+gx{kl; zbv_-)25}dd6s@VC6qa`jvwQ4g9DNp}>(HiP^Y-Pi*c8Rsz?su?|XPhT%Z`j03W^i>8yKCfC zOF6&$4BcQ1Uo(+zc9XfGKF{YJ2M#~FPd zUTuFf0)dtjgfB@6aIbqleL16(kc^%ieaCu8pL~^ zw$FasT(jFpxQ?p!M!KI_f#_uMih0d%u667p==O^CzFIq4= zw2G%^_K!AOEwSUeG`LyBbeneTTI!e^hFrklP2OG3VOxZJ$vdD=aLzhPc=BMJ!bN6f zyz$27Mhrh^vwZ$BS`mDC>#Qrw?#@b;)HGjy;MJ9qGGJ9(vM9~1Ia!oo#hN`?nMqru zWlLVz6q_+s9Gt3pFngFl?Rk!DaoVW}L!jDR9|7f?En7cwZsBmX#nO)4IWelFeq*6U zTQ<<3bC_?R+{Vi{-im$9s&h-R| z|3RaJwG6&XkzU8@{7S8~T%C_{OAAL-+H~z6)M#g=GPt(PcRmkRrS?Xcy%C*_qI#{P zJ9>4IEk02cC+jsi zkfs)|N52+G_Y$h~MLft1?^mwP{N%3YOCds2tsVIQ3YTaE6I&7pQ=!pOw!xWfx1YrR zC@zV;YrvA2Ub>1H)Aj5U<~0}wzjI)aJuK4Js)d(R?)(SQdg%!JxtfXwLL0SOPcHZed#Z?bk2VE`}4C8FKNMccdWw}ouaKI zF+3u5_)ChfP29l)T4v6)wA#EUZOs|^q^u&>f@1NXm_^cOw1BB^j_vy!X0-;wNJiNO z`_nNsBnJ>SPi6JrAb5`h)1Vg zU>3!?I;dl;DzlP+qqjf?BW#f+FX0DEs5E7dsJ}&(LdF*>m+nQVgYiwc<`+YP1D41b z9|s+IUQiZhb-p|vA52|rZgqRSU%1j|8Tmp(%2ySo8yA_(y*|SQ9*}wl-`4rM%PF*a z-OTMj`E7^k+F@?yV)Tdtd}*vL>({H`+ZW4KIjjAiZtjd4HDt(74eD2E#^80Q_Sv%2 z4loqD04=to=6b62#(J{JPrVjm7ho-#hYv?hM#jc>9m_bGYb*WVe?XR7B%~iZXns#{ zFsY|_&|@7!O=WWCSBDM%Y<2!Tjh3g#gCM;_)sTxkYO!VOI?Vm4=E5>?m}CRqQoBs@ zl?itI^JngSdDi1;fJmd!;3{Cxj{`=~+8`l=7!jGsCD8_2 z!7uY2U}hqB5Wk>%O-OgO8{jJSP6rn zsh&;hSCU$9xE~f}sR%WU?R`}bDXf0YD>3&qhDpDlEm8O0W-%%75-3u%KC|=|0Wo5{ zZLeV!mLUiaz7!@JlDLsies&CfzJ@5sZn+|4Fr1urfgQVCLQ z3ty|C${7+wv3(#AFBg{SqV}#AZ)B2jAvtq`$qqlw5CMyg7<*s?oUz`~Bz=(79bDuJ zv8nGx7l3~-< z8JG1`RZCcd^2< zCZnNYUMT+vgH&iY`AXR@Doo>UIfS(4vBPsgI{H0Zu_w?8SrA(Jg^~~tcAd~Dt><0V zwkfH74Z-_0T7-djNBuCFd`2B2LSaGNZQWO$v!MJq$M(;4Du8$l zr%3A1x`0*iyD!`h@DACSU$?ei#dQ%a0$v<45f{Q;9!c!bbqg0xDt&9lwSU|N?(u9s zA=zCjdfiHilUhSVl!A8fGzAz1@dgXE;+Ic`_F_{%v)UN2W-NR!lOZyMFNVVxBgRrP z=(QSd493hdKOo=wG_~yk=^OuJCC!tu*Gq|sCvm*2^Hws6xo1Q(N?BBSJ62O{8iXB# z^!?|dIGR(^F`NvviR3aA{oaWNtkHwoE#qoEi&PM;hc=*D=8I6Zx(9*~2Kr|V-+>*U z1?5D9DY#;1PpLpbwGNcUEQ2WMEwE6ka7unH+i3hnQYZ}x!uL_)8QA{3E{TuWE7(KI z3{HxUbLEe52{sYtQN^s^kH*5=(CpNR#t44?!nSCVNIGR?Hbew-4kf(;WyL4gDkR0q zlgldykDLxn1211!1(a`@ZY`~-gN=yjiy)pF@IJe65`k*F*Q zXM@Z3DOtv-Wv&Yre&UVBqr!`U9t0IYqk44tfV9GjM>wD`O_3SxC>;3Z`@q$6V?Vf< zp5O-b1`6CW>MYlWic{CU9yw$%36Ma3_0~pI&X>HDSrMKl^^h~HtAYKv9;+aNptG{% zhogza>nR6a7)=NO5fedJ%-pj&`VQ*Qr)BozI_CK{?98&_#FaiEc}GNMxgxh9i>Iza zAJPV5crcEn*ee&MifBlQF%W}p{a}ogy^*q0NUjGO6<~dQL?m}qXD3AnuF?q5rNOU; z&qH!vGSnqEKV^q?yxFHcVHplEFlSdWvu4}6rKB48*IyG05H7RHMx+2P%eR5}# zYh62?-l|DsKFqjqz|$wxJ!N7F5w$os;KZ6Pr%q0P=h2jg)82r`onCe4)RYR$$z^v9 zf>zAL`=Xi`+9En0Aj58kbrx3wD>fy&bCM8jSUs2V0(>Bll?bGSJVHmN(!-s~SXNOG zbPe}mrUA3KyEkylZ79wdsE-rJBQB=*aR=SR?_gaC==Mo2U^*kGile#V##S?EKJg|F zvo=2`P#0nAaF$vGhi#3#g2F%G^#1nt#~Jk9GZ@(jI_6_7T4lYGScj(jvY=`VLh4~5 z@>2mN_lv=IMlHDw37JrvQLk2+HbHzK)(@vn>75hN%T2ad8Utc_Q+wyiGp*ES(CHHl zju8?}QKp<)yskgU47{3iTQgi;_@>i$k)%J4aau(^o7e87;E3X8wAtU8XK>!^5*C1S zW+FE7mKYGrelr&DOm#6qSqrIvUN$B1{Zz|Bf|K-y#~99ZJ5dA2C?T96b(^?$y(C6~ zY)S>^?i%^f%|-vYaIy))OwmHEBN&@3B-y}dgff2!;f}(wnHAbzYv$vC^hF&Qg5u8Y%&d_)2RcFsz!|eD2Mj z!L|e;MJy9H3V2Ci_>r+fagmnHy;WGA7Vbk~J3iykuDuKR1>Sez9W(}`y={sZW(@`7 z9~x*nW8ZEPYy*#sK3^D;BUEwEiJ0puX3lx;Xyd$E2YEQKj?7yjL6Y}tXEh}p()+$0 z@ORTd!$uR5rAM9A(bk53P#KnZm}EO08MRt$(l|IB@#tK|P)?Y-v4}s4EL@DVVqL*7 zX3bRR^qGuUV9BNjEB8>ua@Q~8dHIghHeF9&WiPu`k`TdRLUWGLFmzFW@KK^#mNiq> zsxehPp^*_U-4;u=l){B^VQ}9P?0AW{u<{w(c9B#!s4zpK4KrF5ip5$0Qr%~@h|(xb z?#-Vz7u0e2%fGe4$!2w!anf!ii%maRQX=%Xavv`>ZLs6Dvsj~O!b#|m8En?L`{d2m z8^`g|)P#6^p6O05bn$ox`vHtb3nu3W%v3w}F_ygbdYhsq7~AqayKjNHKSlmn6t*1; z6{4n7p}F)eM@tCy0087QaI>zL=kPhv{xuNsK(SLW+ztW$gs>w0D2_RAU;xm_}SKo%xmni`yB%Jda$Wd_(-Ev}-2MIQ zG}I#KTX{1Q>?VH{a0^At8{ff-MudY8sPeIu=FGH!%Czr` z1z@;VgrmfWxbSGe5i-n`#pueKH-q$pDUjspZP==9@Vn!~cQ$w7ykE(^A$|Rkzl5!y z{F@klTfPBcd*%-pbt3fggwb6^G0MuPHGBfc{=^3A^qFp^3`QY!vvm=fQko`t}X!)6k)e`d5Q*B8S@o*d-xoq&Y~ z<`;1Z@ChF%gmu7xhHc@@!`(<)z2P2Uxjqn@cgkH*&_DV^zwVm4qLSc`3 z;CG!dbnDb7nx?5sl+gUSKICVAX6RV{lw)l!^W9r~qCrY0efAuWVPKd}p?l8U28fnx zC3?f<5x-LvcaN5bPb+#wBg17#%VIun!jRM(%`gTei#Vbm+*{#Z5sAG?`dm<%)`|0l z;d}}yl}V04NPUVUoH=9^j)Oe9OBmJPO-ifCCWBC_TO-|ynn`HeVW*FZc6P1-QwV=4 z*_%t&KrHO+hnYBmZgzCTU>}VhZ+bHY)>x;zLqg?U|v5AsrUdVz{@2fL4J)2Te2%f^q#5Loa)j_h-m``1eStmz_qhdgmp zfdVrgBf(+%saGeStwKz$Sb%0u6Rl#B%+Opda*WTO+#r$U)qB2{(_)*JsYRP+06zWu znXk+f(Z8;cX!&x*%E18uC8z*^(0{T*;^d%jZu9#ZNtC+2<8OPzZ<}AiJP(VmK2&}d z4aayXG-MB-$|U_cAm@0T`~j9mv>zjvM1bn~;Q=cLpR-eLin$IFMTf>^%d?#!i!D1! zu37xh-H_E+vRU>Q^Q;{#`jvXh3(xfAbk7q01Ajk&2ebR^zCZgkQP4=n_2-wklZZ#SQ-Vn_NqP9!@hmeom3@;5-zlNXFJml%Bs&tT65e*iDVjDowkK#2C-3W+vykcZdY0@rA_!*b zGK&2x_^qPiH$;i|1p+)|JOl=BEd^oPRO*&OQy&LxEanohJomKED$Y#SX4L(Q4bGY& z>}!lSP}!RFoHWv9zyxnO)ok#a0_|(o2>O*BOmx?V1S9=Kn7Wn^@x z(ASHMM;;73nJf4XV;$#>s!?UheJtk7nE2>IYcA6)sq!ThvFV$1e1jS8sC28;k$W6A zllu3fVR7nYP04a5&2iTFh6?52AX;TIRWZ%1;&Rob(l$Nv?oYR#NORbpu*b+D!DlqP zm82`lPFbpDRo6u{^(^^iO695!KEVyqo9>W_ef)my zo$L{Tvd^($G}!D(*~@zS=ENsp5INJ(@)@{c9)e#Bua~{Hq=*hXuOtv)pah?(=uPo97IyLy#^sXQZ%ETG*>Ffk*}`pC2;LI=x6K25J4e`fOgS&@*p8Z1Uk$ zS0Xj72xR%x|17bb%p&r)SCodCW9wQA0F+}~M(OkNcXxGi@`rXiY}ZG?w4=U%eclQZ6ws%^=mg7asv5Q7BMGP?=>_sk<4{@EY%6Z?aTySEO@PoOJ~)~3 zBBr^7_W6Wofp`#imRplPgko&s~}oIbP1dlvt{uqDfufSFWQ%2(4wgc?81#M0o`;f1BvJ zwK=Ovm20q#vLK1-re1Xbd;;?eO+c+60A0$2l~+Zs76KrBW@_jjZL9Td5M3Pn_QW*P ze(m90``djJ{?ueB>fdcQ`H3qOC8$~5cfG5@!Ie)J43{TybGce@wMMZP{3aVQ`^ z0iF8e!+tczYIGy^sUngDG8hHNxJwV$#25?`LfsqGiUGznp+pd5&kG@OP$OQ-kbO~T z7ARtmA6O6gm{gd9MQkAs8Qj#LqqUcz=T*yUDvJKY*kW)yiDnZZ_7(puj4@mCq&5&C z)4c#ea3=_t@#4eLX^K7vlK*;8oxPtJWl46+-OL-S?P#e%>+Sf)cB8xYDLND zNkoK&>Zp@yH5OW-_SB7N{FJj6<-&NqS5*TI7J_75MV%N{d7{?g$QX)3m4HlkW8?FY z5gdKQ?aJ3Ao)08sD>^2u`>j-w6^oi=W5%+^nV1m@7t)Ct{2J4yO7mc>*$8)rwh!LK zHijBFSshnj(NJP9 z!V=Hy>gbXh{Z_SAG@3=ZotzA9#Y^Zqf>5n9)@|YpiS8pa$7}^+Lvy+c`G8k@FYT`E z5?1W2a3J`E-1nKtoO_+Hpp%Mx!vGxx_tK5r~hW z8FnSH!4K6{n>lpDUClBxsVV%9Ds#~EIa_eS>L8gFQWzR4xpb zeZ=nsq&W^x!H-7bEqgK3oLY#PU`sdUdVCHuO;)&_L(%yJH_!11i@ZiI2gJxZvFkJ-2U@JAaH|ALXIq^@Ry+s*8xcXsx?U>*4mQ#WX7a{5^+}2*};Uf;^B8SPGiO%B}Ob=AH zfi%RIm3{gLyYjk6NiOc2e_ssO{QqjP%54|yi6;|(E(my#k9Ma`( z!N1zWWHQYc1a^W`zH;Vqo73m9`34WP(;#H@`gdcU3h2Z8x>Z+ul{pCqU zhAJ?UMoznwu$4G_6VhNg08&C>%~4pBBw}Lu0fx>X7A^;nEA~CR^izjVm4@KTeM!_S zr}1Mp=CMW;M9QLB&E+>o@voYPDcOVaBmOVny7KEpyL{;B-iLHaw7)}HEapim)+cQ( zRh0W~6j#Tvo)##_u#fY$NjDho&9T;x7;g+)s1|=*{S_~zN?)dBu0y7tRw8HCN@<5^ zKHFB6M!hs<(p7@3q%7K4fjzdrn5joyOyWR3UFfbCKxy&#Gg@`IRB)W z{l8TFD^97Z0`&*8`Ty{me}4UbCpQ0x-&2764e|Wnr0`e6zw?`auKzp1`44@S|C9dT zdCtGt{Le(^-;4_U%l{Lbzww>_xbgq@% literal 0 HcmV?d00001 diff --git a/datasets/subjqa/dummy/restaurants/1.1.0/dummy_data.zip b/datasets/subjqa/dummy/restaurants/1.1.0/dummy_data.zip new file mode 100644 index 0000000000000000000000000000000000000000..48a0d552275c6be8543d17d662da21e5c1fa47bc GIT binary patch literal 9349 zcmbuF1ymeq_Vybm3GUWFa3{FC2RgXByLaOfBsc_Ue30M-cZcB6xI2Up+=B-Qe3O}# zng6%5XV32K(_L?!Q}=n^TV3a^zFj{RMFd1Vz|Y8aEL8h#@VAc_ziurcj*i~O7N%fR zc9p+8kl@b;I-0tJE#26BY+e3M4$WWYs6))`R3%t{SNToh`EL{Aclz`6m;q#eoZcT? z|E%!)9=oQ-GXR2qSfa_#H^l1&01ojA4gmPA_P;KT??0U8-%Q*r-NB|1H&Z9D`@gAO z^uHH%cX6=&gEs1KUz!dHl8`wP0Kkd^0LcE&+TdSnU^933IMyCt+X4QEx;CIKGwbjlaSW3dL~b$-co3N7iy z+k&hF_IiZBxTO=0K#3~|+zQW1ryVru6tY=%%iK~%3F^j}l$BGq-B^v6By89vL7To+ zBtTiyky{cG;RJU9p==mo={K6H&Q@LlA?MxWiwouAxv~>mo`ak7y$YGmd=+fw%WT=cu|aiJ*dePGqdap6tq&onad4|p8x-wr2CrC`qn!g4-uPxaF1$|n^3A}1H_wTAm(OwVs< zB;r~3uok}l_~`R^G;%*@HJ9}$?UJwmL0$fRrbP)PfSL;*UO=G?+^E9<;Z)>{cBs=E zTi9jvsZmH%eEAH+w{awlrLKF;pydSs6nr6hs-%g!J6F!=ffF9pNTL_DyBiJBmfXv zs0cfy6}5)y9)E;R1}%?7m+au(%$wPAcQKzcLQGQ9K9e;NIP%D;L<)AfI2%T79YFs> za7A0)Xb+#nr6JT!1^Fe)ytfpsQ)SS*mHw(x&=QwM4355U0#~~#OWf?P79dzn+2&;4)P?b?u=8MZi&?jm-mN@Jdj^Z@KCp3srXn5|@c!bi$g-@q6JLp(wy z6yVzPL?STuqXz802BKR$cQe^Wi@XMzlE0b<(C-jCZUUj3`4!@YZItBXPNvVLEQx!= zu1ymN&fn#c1Ebwv&EGq;T7*-I1>fEyeIP~E(?YvLpp2yxItKBANy!>kiBPzY<$o05 z?Mx}tB39R8H($jERAg|TXC@qkWbX}49z74^b}N=;sp&O#8{7oKwO{izPJGBB71lK* z8wKxIX1P-IVrE=CuJ8-6fZ@)B-M?a{WjMWe;);5G)>t~wW&k`UJ35Q7>~@G5#XKf? zl}B?i<~Df*QEv1iVx~4jO}-3sa(&AjMf-$WnC;raFXgE;6p}tCp1%78YD0{-lQGNWMslrTC*)D9D`;)H`S@vXJKK1 zcDaW}uvCItxl0(cJBCnfKTtDsbP4rra(u_MTJWpMP~?!iAH39iBrRTe2=G}`uDaic zh;^J4FFg7B03UAdUi|&O^ZPQcQ9r#KjR^oiDF6VO|9Kg=u=M!3iYM!?xUIT4=-d6wSTmwtDS86S|O{@Z82k$Z6xaqm^Xvg#K5|n=i#V z4qOHah5^l@Cu|dKgWGe1VxoIhm_>I16rBz=mNt-G$Xb}iajX-xr^R4jP#Jtc9y|Sh z;IKg<_}lwMApuzG61df%V#l^KBLn6o%ri1)%fVY`%Q2Ih97s0sN z8hV9|+aMa*3nJ>9o2Hqv=#yP`sL_+#)uO@bldDGy$Y@p2JFqOI)Q1Hs&{x`!teRft z(4keKo(gf%bi7JvJ&=sHKGZ9lze&P5)I4YtT%qhjiZ8GNLBJ?t*uMJi)72_Lg@(EK zY<8!D4jja+1EBtNS5dz8t%Eh$`{{-Xg;^8?qPgL@1A}`HXi5F-5`q@fbMxE6SbNPz z*K}rxy(|UBr*A)QZHm4qnEKnu9dw|#+GbYhLsL7JNs?FE1C~nBdCe7bSW2fc=6m#w zyRnFpf07@#)kZi0DB)3pPoPY80wGlEi6f)j-D;OX4_}Y3EmMyoz z{rL3y8|$c6gCHwNrBa2Mt!4N&erTYc+E-K1LXHqX)S|Yy9EP5E>)=j9qwUbC7(>36*G^|lRq@qQyC!xay<&j*b@Oh+V}OqrwR|E8;^HS!lp88-5#upb&kz^rSCI9;>gsjPW1^fo zvZBpi)fH3c7L;hg+t}HIG|I)}j8gdIHjr;uNijr>`28w(P!=#0mVGSDzy#HNjb#GE zQ)|fHy}eaz*jsx0ELfPURAYJ}!XZ&Vs-q#&%!)}5fSM@Yf>@>` zxD-RXFFfIHbo;n9kpU}3hWs3PE=$f>bv;acs_o5aDZmpi;5E_~Xet@JpjPO>eU1n6 z()Xa_S=UDlnws+ISTh7pJAvoqNw}PE7TWLX|p7Vv2D}mGSrR^dLMd6LQF!S z6qM>$;1t6?iHO9X>=e=9`#d`-7NVh6Xu%Z;TsgX+%gEbZ z4~1tq<}Fwx3kUQ4E;ajZNU9Q8PclM5AIWNxMrtv5-Q8%l<;hAUaBn>N8f|zWbS4&#|p*k&EdUC*mCNU%oVwi-NJ7qvV!tdK}&|3V-VRye+3zmbgUYCn9QHqYUW z)+Z6a!IL5p^KE#dFA{0DA4wK&2z#42PIOz=nx>{s5XlFlcMd+ z6nH|OY8SvzTXY#XgB2SJAmUtVJ(*mrx_VtuONBi=a9RK!Mj$Vs~aE16Ip% zTFL_kk)kS;t1TlYO>R}u5$%+uUnMxXeovFEZa6~t(oQoZbIi4f00_-75>roMPkbsS zlmd?{=L!YkWB<`uPeHdko_6(wwPRnNtYQEK+(PNti?~GQ zebSY$dMuVKGT*1xG4^fmE7UM(T}bcB1gMQ_BVFzs6-=sg0ol*@y;{h2>-M+6lbc%A zcnd!m)5(Xv9rlNdeO}u!E_11EX}D0{^<3CSe+Y_lVBxu{M`4^L&?H#yF}kw`hqBf-!^n#S+G5duwb7?zNp}g9N^sK$?H6Z6BYi3- zMU3O=M`375Uf+p3D2NJ;(ob0e;yu3_Q*RzCUhc?R&qES3SG9ED&!wp-{g)RL&EPGk zzEmZaik5se=jH_pu0xAn?CP>BD#>on(yMzmr90_kCvrlnhpZp5{m0(LMZX-jMO>^u zYU0z)Nno)DOFj3M@>+v2=oR-MpM?m0Q#_#G)37+CED6N{)`Zy*I_<+S5^w|ztuWN3 zILoKKQhRvEu|0t}sZ$&$WA&_uB>EERG^=;}-L1Xx;x%A&7*0=rmtH{<|I4s?_*{5H zkwZD_8tO?e99paddA+OusTjA;NvQ5}iX=vDAj5=nzRbnn~ft1ilbZ}G>|EKweAEX{I zrI#WAWhr#usD(HOWY-naDH9bLW%yAwb3cxUwW!r3bd zX#^OHFy5_W>M2omXj)bCToqh_?=ZcSNE;qWmWcoHgL53(bv6K|; z!gqKxk~!Y8!%XJo>`?n{iq}B}QlPlZD}asIHw7?2lP(%SFlz#hl9B#k^)|?a^jcNQ zGF`%k1QJc?+%<@dKqlS8lPMQ%Nv+%Tu`Ujo3M;Mh9lNu>{rphRJ44TGQh+D&dR3Rs zfPZ+rJ4{e!L?4?Y{z}6yj6%pK+H8e$qT}w{R)gRpJp7y4*b7XvZs3tW=ElbL^Cq!x zcv5_9WM>a`3wRIW=@Mq81KzP#H<|mrw+MEQ6FCkXKo}4Seo;#|v>W#f z;L6$g9EI-4+BQ`1x@7fj-}$TR-pN@HG#<4;)fI~(vAt++#QuDWsEy^BXJIVeuH-cH zlZ;?~8~FQL(xRZdvO~{0!35KGi679IZ7;WxM5Z^ys^RXJ-KL>k-_IuvqO$7SMAT4E zX&`QqEUq5erzK2j2o?1s9MBf=D0(7ndVd=4+!zme2W^x^+~+ua9F}c0Aq6UVxO30eY$)vKb=BaWF@=MgG9$DxJ z(H=*7XzscWA`unByel`@q9UaTHbfF(sD0ce_H4tp0op1OpDCSQ0$Nelp?Fy4uP3mhrfLK@NF$2^QgO)%!(q^8WU7152$HC5SPg!2FfSAz`6IL0XU7P zbJkDrf^j0&Pl$6(rBO6E4KXgbfvU9O)*7{rb_;l75!vw9_HH?MGgA`I^J&a=DAhPM zJh2lf^y~D_R^ZJsy;vdaXPgJy2h>`AjEWewr=vb433|JF9b6h$X}yfIIiEHp*g85s z^v^Q(dXgRGH(GuTf$QK*laXr)!q+vzY|91=I_5e zy-7myjIXv4w4CW)W1gk!e8DzqGh)OM{(Y zX6^x;S*gaAP2kfaXc)7r4#h9NxCAqPT20G3I8Q>}Y+rK{qvcBT=L|TX2o3aoEvol4 zxHa)P;KdD}ap}e5m)`TERkMLkE+wt*M3n?%{Q2d1VQZZ1{mI_f`t%LB6U=+(s6IuU zY6vE+S2Gk7TlN?;0O?8PBV*D^8R}!Z*!YAYo0kM6#AG%OYbaVp?f%^t1=<&UH>x-Hp|pqtw>%K)h=@L?2cn-#J`hUc6Q^qG888 z3J7_4L7e`a9_GAfCO|-$%hUR{$@#;w*|{xu)1*T==iWMc$CK=j+x$Bq%~w07I{oi! z$LLM9839^mSC+&4qjg);N&R^LY<)#^?y+Pn`0Hfi%q7x-Av^A>85WI0_frWz7e4%No2MObG{s5+%h-AC zvzqkQY@Pt~xnH~Bpy6_uZUHf`F@uhZZIa2x1SORNE;o9y$R%|pILK=*vK}^QKNZeR zd+v{JdMjL9XiOAGW{j(h)cLbM5&pT`AJ_Vd%k>NZXvPKr-4kVkJbLQ z-{;5i&#CI#qE*1ppEKN4A6CdDw#KFZW?2i1enBISP4Mg$fI;{1PGqfsEjz1XI2sfL zXuZGOp8?@Q{H-5JEQazpb$}IH7ara= zX?5e7yuKH;(E9nMrx88fDli31J|;^)Xx~;`;9&nujKzcJ*7tlx$I=amBkbHD*3jFQ zleSa^)(W^cigjk2@e}_3ZE$dfZ{c(rlD6h#$|=3%V{0zYa`Hw}DvKdPC`FfArb`#8R^=PvO~vZzhPv_i9!H$7-m2x}~vZ*;KJ}{9 z>3dpraMp9DS`y~O+rv*aq~7LU!Rgp{Mk@Il;Ikm@$e2pPtnB>RlT&+ozXo}Q9JW>` zRVq0dtb_;pr0C{XOasqU!%7lLgKiz3{b*FXANoGWNOcx1vJ@htja~gRYPh^6EU=6- zSVtaD$!g%VfF}#nJGCf>b2y8wtxoROdwiF9m%E4ekK>1f^NX?jiKf)kY+B_&`U#`a z-Dv62)cn+mHy!kZjZxR6xmc>#vDkM=+J;WK@fw%p#p||aj#_ZQWVtHe zyt<=!KCOjm4dwljXY6=XJUHUK8{OU_E0LXR;9yKS6`8!UV3OO^(KJMh(2`7dK9=X^ z#EKd~a~KlZx&PrFnX$Ot>zwTqRR?JsL85Z|;aYg@9E zp2DBV=)@Y-l z4HB;2JilK_4N(?S{ZJ{g|6bBI=)gc`YiL=g$%BfWrwb{cOWnFwz(%Hzu_nc6Sd(z& zX-S#iWlhBlmA6MEZE!H^Pz;`AsyQuUS$KnoUW{G8Ev%0O)Q@7KHJxu@gtR0Z3xbfo zJ~aU1^u^(5>l_{@`Zj+m3@+wG7OZ`DrN;!~d+Dt|=3>;h8)|M;D+W>T-sg=P{iT$% zKo1f1+7Oa4`GR>Dm-kZYvbAV6hXJE_4=?SFyPu+h-9>vvgtxCK)27e^OG%613h;9z zfhxm(xAPgAX;9BHxrwz65%dS?8=?!_vMEoy{aDKZ@VL(H*S=`{cSz9K-m%a*!S
    V0QpYjM(fR{*02V4F6iK zGM+NLt|b1piHRLl0gVNqwr9IbpkVbV5jOhMRJWDdN(G+cnc)$x&t%tf7 z8`lo6ZJh7j$VX0;*%==Q{c(ihcul%5CquM{y8E)_5yWbnN9+`pcFaIldFdzu4Ii4q z=K%_L7N81ze<^g-0!|X$#8{hv4?){X=7vx%MR&xR9Fki2Y-a}J>Be3?U&rWlW!7kY z>r>PE@Y)Lw9=OU7A_EmEF ztY9*eI!tDesP$N6uUG(FBIc<|4isAMv|3JqZ zPUE)g{;GBfWgF(B;+GW+E-3<-PF+6!RQqx|3Legsh$8B)w#rB{IaHt z3!8aemBaBS`lH-Sai}rgPLVIKy)7z&yBi|o@q|ovjuj)9E%-J$*djEs)ijb6 z*f|nExP)AZ`+2h3MGjbn9*2fYrG(|P!X`1;wFRdG13CHerB~M;H@XG#Si_`M$UkE+ z#Pqt(1GB5(zG4VGAW?;MN2al*e)zT*I$&&;UN-j%Jum*f#$G%^Bdlx#Op+!_% z`eqeZGDv-tTcKe1;^Vj`TXesQX*@}pU&ink7Xjl^OHri(^(+uwbYT+on?c!jo_sMA zGhSz0fzgnq+%CfLhmc?_{p6~pDFrcz;ge8Y?bDVuH2k=??MtEK_^*Gmiy#!{I7f^C zQo}QymZOSf+i0VZDsu4wEcu>=ws8o0MK{o^1br-;LG>sVC;OXK8TBtCdYj8pa^=}S zi16_s7{waJ8l#fi5o)TH-`$j6l-~!nOoyekor$I z4-yCBaH#2cMKk zQt52<69Y0gfkkFV?kkqCsYgS?AO2o983J%{)veA>%^Z-#P9Jlk@45z z9}@q-D*QY1pNz!s@l(U_*Wtg`_%rYDQ;{6uub2shl>9oD3E?9DE>cs;;&D` zPk!R}_*nwtufWBBQo_HaDgN5zCr|ND6W#yb`ZLiHYT?5#Y-OlD%+m{=1`?EG`ix%~cnZr!?fb#?dd zs=c0Pt;XtFyB-B;FmN=$Z&qMatn~ZfF9+CPXJZ!|8xK8W17`yUg+HB$@y`=&44j-z z9O*qR?Ej4q>7V*2yBJz33eo+6@;d^|?*WO~{&RRV0Nj5J?;lwI8R3sB22~Yk0GQ+F z1bydUyNf$400jIN1OWIw_didK<39}ZFC@;67WM|lt`<&qj{hci{{J4-$==$++39cC zAAWympg)50vA_X<3M?ozIyp-O0zVz7ZMyN}J?X^`R`92Q1U#XAy zA}$Pu;+OBLi@4#kPmbOPba0AkZUP6PoKrn^jw@M9lbOZ{x>G|^AuTCni;q?6n?W*# zsoQ7Nz)N|`?^d-b;PwHvy-F5~Erv$b9cd{R9Xb?kRZ{KL%IT)C$tse-+D4@X)pGDr zvXW!N^lIc&6%!3I3!qx1Ds?4qGRw~UUJeN9A3PNitweVHzFLyDeazfK|gHFMZDU4n~O%Te;v$lc02Y7CjmJhL? zv7r$4JSB^}9Vkn4&713!_hJI8hiKzj=(8n}r$_Te8QJ}DQy)gQZL|0=AEt+EQ4q9c zB@t6^Ic7sOKKnNy<^`3?Zm{ahviJvMBPu9e;fXb>H_N_b`9r`a+jM@01VP}oHkz$( z`&J*U#X~@e9oy~|6>uCb%ZrXaH45lSP69$HXeml0Bn4Ej?=Bata#uusz2R1wC6N%3 zh-FB2WC`IZ8Tvu5Nw&FlJLxJBgo6iZeuJ=GXfjPuM1nwdvb9M0m352aQB=#6RVIrh{52!#dPPrNLQ}0XR#W7d^Iu0q!Q;Z&A z3EHk37|CiEVZN%$Ui>g-kOoQ2!v_`kK2(QF>myf3R}_VHyRk(Z4M_-KaYr^zNi2?T`hJSs-Qv_8}e4)T{N~{;iq>( zIpk(_wnfxCf+F(CFg|-wXfvZQP=#(DU_0h3j|H+&5jj4b7QC{%)BEMk`R?npeCq|l z#g<(Et&WUtb))DPZoN0IcIRNTT~GF!-w#`TZC$Pd4r-JXZte{YhNWxt)2T*TdS2L4 z6fcXe#DQd0yVfc6NY||CQ=>Wu(zHvFV!9L~TKk$gHR^^vqlcF93Ed_O730+0DF+tG zOw?1y zh44z=2Tfol>SOG@9=P3V^!4R=ePKXB`I?nQl1<=Z&l+R**J}Ak@u_#AkrV$_c*OAm z0I~nM_!yhG{#JXUH7}gj*il}6{U9Z=|7?a!l!HzJ|Ht74|xc!FUH zUst`%ZZ;^-@@L$K)WKl$?o)yFQ1VxAy^L!wcx@)tF#qUeSwFwMp z7F|oPyWxA}Zu-FW(u*!Xe7curyv6q=rRmn4#IQ&y{%G1wKfB=BGD}X=urf-^Kt5;8 z%r?iGdHU11eO8)lB4n|ujdMBsve=AT#m0`_2q{+V;;y(K z`3P_$EkAQ(VVrz-Z>1SVkwCV_IoP4l)?N;?dO?M&b7e5^Wpn5Q{+Px`1d=ld7A?k;b?PzEQL>3uNo>~b_n{6Ox^C{+_6sx0S_qwe?;vr;cB zQH7KMCVIwdS)rZE`Nnlwi|15{>VmPhDtl)1W*V~BKrMBIJysXDhUnVQoJAWITV$$K ztbN$qzW9q)J!~M3>ksj&`56LSoPy9B0ZorJ!n|S^^2@I>rNw<)F3)g%d=*tUePuVC zPmc%ZClirYx8KJ>LWPj%b}Eil_s4KQ66Gmdpi(_a#Q5hP+~fI2ewlu>x~LTF6BNW# z2++_K=T_-Vo{`&({Xwberu+ITy}K)b_(n*elHd3NtqySsrBOes(ChZ5fD^?YM)V zAB$P$`!GNaLgtFYGpyF`479T_PUH$B6{tOhFTy8QZ+X`_k#2c%X= zeudSYVG|F)a&~JUX>1FFrTbYXrnN2So`umE?pD+jPkiUMoHdeaPG4=wwLfybG%h3D8M0p0Grxuz9gZU$74Vk#YckO*=w1-NxViU9_2 zUhkm;{Z2|!jDvfP%+VzAtYKoOF_Ka|k&m58U(g7k@(Y|9Esf%odR)d6YPT+}8lw=EE;JtMOLp3@{|n2Y>~9kFXizD??s2 zPnGC-tlUg!sY*Ozcu(7{ewUt`k#>@B)yLmuXaoTrriE_&Aj|(CVe~EonH*U>%s? zj?TB+y8(HBam7Rvogy7wMUJMBytcX-WOOFKCe?Bpo0rYPg${q@L&b%Dq#g#v>+DI{ z)}k9KsreBU2JR5gI4)-~lt;9WudJIVCSb-2J9Zr;-eTnKRQca5iQ-X>0LwSJbvUIN z**jXqas5(|`$w+}XB+#W_o)}>q%yf5lHn)O5PQ1uB8qJDcJplCq~i&~!c+Bq4_fH- zqfg`4h3aBT)+M8ng@rYASGPJNA|V76Ben*S1J;D?rs0c3ov-D-aVAyJo!w*b)(IVh zs6cgm2J%ouPNnDBe~)O(;yk8=BS{a%7913FxpJb9yF^2{uoPueHyaGNGPtpu;}W|w zj45o3<~n52$TM#hnR)v*9}o+>I)JBvHajQBt;YX@8_C|3Yp`RU9F5a;j{@N~86lX) zu`5*~>#-fej}1}XP z287iQ{b8MFlEv2{=8@ z5Ir8`fIXeB&bMzH6C0e+n#2l!66Q$6W;IP;+LT2~#4>ArV^;yT-)svQd z=@d&-_omZlD)Tm7HnFQ5q`_YKu-4!piU)L;lJfd~^A007<}~$-SJ`U4-X{}$78FW_ zndZm|@g3G&VIxf@epKYiF)J7~+x0?G6npqE4r_Q0 zlQS%_Es`MDO_YUz``TP}9p5%GP+vOyR&=x4u|Y=vzNfOX!1F_m)}PLzdG=6t>5=JZ z1xc@289ALj$$5tDm3)|m=yzyC(bZ!$edF97qY>hud7VA}o1aVzDz>nZT|kXSRZjQ{ zqGfG8e|{cAA_8|Ygk9VC4iaJcO{Nd8NDZQ+d?O_)F=_XJGB1cQ9JG%J+roMC7%64+ zD=}F8xLe=7Cn?`HQY$DY(Y6~J+x5(}cO#?giJVc29!yS<%oz*S=NYk&sFp0xj z&~+#Z5Y;8pS?AQA&HqEi4ywIL|1tJpO&D6S2$8Pi;NCkS)qpi&9>$yU7NvV3&eF&3 zq<&z_g2tk&N3610umBDq&7S2+);VI2oIW8VzO+BddpAqJc)&ZmY`_shuHXeY>7nuy zUe1u=9OUIggFJx)Vgi&;#RQ6Cl~0Lc=Q@98HGRsld-~z-qq~Pg{KQ8VmNE6+M_Qar zZynOe$N|z!hNU^;Z2HAx>Rye~F&+8A6~lIo+Ef|-%$6^M+c$<~D8^3>4}PT#iI0w4 zayi{%HlNBJsslU89YB?K@8as|Jn>LU-U8Uz^uSPKP@^`9RcetdjE1BE_H~zW9cS25>`wmP>m$UH_DF zQ8*G@yKW&H-fk7G$F$lVn8q&8l+(UDYs&9(Le5GPnW&qW{v23{D$MGXH#CwfgGX(1 zEk405eSsRAb)xb0D?L={tZ~wkZckUY$2ExIJIC$u!A51}eXT&(qg?PoxgMIQu$v;2 zt`v^ZNJq{x-JJ;E5jTn?d^7ZRScc-Q&yWKnx;=5dK^o!o&UR25J07pJh!AU0Z4+3 z=#9vXhLq6su>LZwZE%-_JD77n5sNxQe8bD(qVlm3uQIjCF+GC79Zpb$ds!NjKb7pg z^lVsA?@O8W{}?t)aDtX)BdxBz7rz}ecjg)0iyk&%4kRv{ks}O+v7D)0FA2sHS|z#I zJBnidKGQLsWN$7442%J!z7|&BFO1x~GRJ4gh};2mlFldSF*)7CD@j!(%xGIzqUoSK zXu1Z3P=Rw~FGn<3&qbcRJee};*P^l|r*ge9?<&1g)%JJWeFBTMv7?YKw36$sv*-k- z*DGQ$HXXx}w^0W_sYp+Z*+-n!%q0|5;=NPyS&I@Mt@#;Xx@l)u+-qD?rVb**(V)kg z8P6BqRf-fk=k``ikPhIOzW`Fgtb9#9eZ|Ew??kgaP`J$i$@5FvMYNtt=B{)tJ=ikJ z7%wnK-(zgcZ$*q^Xa>woExI~yJMlI+Y{2IiX(7S1=b>cb)p-^yDD<5Xj$=cvq0B@W zREwR{Sgt$CB+4_8WzWv*l*sY?{i#c(l>4t;JLD%EEb!fU-996!2nSa*DkF9m=EO=q zu?h(gzy~YBmC$Ms+!jq(d|H+Bb|5VqoeT5eq**M;D!z0m4Jb*gJW9O!*+EKb6-sx( zKgyV$xox*aJ0&pPsm3gj*Lde0K0f1OazcmYcx+jq;Y{`L1XMEpDd{ zwfGP}P2kwybyggEVOXv}T08!pCzyctJ?SD#)Bbs2!kd?oCXRWYa(ky)M<%>KPfK4ymSf6#5(P7 zVriaHCw9zfM*g1B1a%2zG=-5UI+3Ehc!$bJbo~7t@Xwv9uRNh82QmPlP4=r>{hvEk zXGa4I+uwUtEi0$B;rd^_>Oc|=`iN>BJwX9J2Z7FJs7-N0i&b&7Kn3I%|9+G&(=7hx z2TNN{9Z-ZOb`6p?*5Tl$o*nYJ{2RL+Zf8b7(_%ZD4HMw~@>7D}O50cu96jzH*sCPJ z7u&=ikpS@$VwW!=9`ga!5XG|Nj3jNN4LFHJSt@&a;{h638 zUi{wjlQxI^eVFHHc-6&FPqVR7?x#3akvU@?oEDeXNti9Z_8M29fDfxSP@SWO){fq6 zEK{sqL+!4jYgD~;sc|fG!7y2MY=^lQxIzDMf1y~LR!zN8j9;5Bb>ZuT-3G7M#eGRk z$rfiKhEFzK!^!}ymsRGGTs4*vvz`upZR1BooUYLKDorReuepb>%)}oLYEq|ocCC5S zx@}rY?hnq~ryI@r$0SXbz8=88!!fa3us6dOTV<@iwEYaRv%X-aRl$DfP8`}|T_#)h zONSBwulXK&3+dPA-(Ae)v5{%{!JSs<&kSkJh;LAu(^qMVh9b!LhmN=k9ihyZex#0}#7~~|T z#RQmm=jrLp57|E%)2!IPdD<+=87^uO%oSyY8bumLRvlJItBy> zG}c$aPTX!4B$mpN=3;EGJGH~7iA~#O8%{G1~F3`@KCbArcU5`o5fflV@n~%MD_fJXB9jPv2Va@ph2n_WP1c zSD^dc^X|6y_#pmc1bp~rx8P^IuXL~O20gucqor=m)xZv9TX*|p<<3<}SG(r^()WoK zxPC3Qfr*9v=I(Y{>v~F#vaa$-^1GxEJ+o_p-N}^+TCbnl;Pi!dcwme8 z-xc|E179YW^aFxDL`A<+;1YTuOv814+LVn0X@H|b+)8|UnAi$E5TxdR&dHA*C*+Na zD~7ttuRwai?X{-C2^l=CF^4RPH_S(nI54Cs-&*Y#Rf4!wy1`q1@fY7%%049)l3fEy z4+5>dJgWi8@FWxc;gQP8`_nD-Jr5=00MN?XB!obaZAHcc?oPty%-=L3J2VED#Ok@;WrA1BaP?d z;7}w|Uhjq9y=i#R}fdj^nX?%1(6bP(6Xk9Ub zY`9TElwi>|auP7_i<5n4ovN`_kE6`pD9;T49xHa?^x00CB*1#Bvj*4b4u7?f%lH8Q z6rIoSDIaT{TO+A7!txqAe-Qh^Te(1^+eq<&G$fjL41ST#L~2K_cD+4~l^Rq+>B-wP>e# zYTL|cd4ilXorS(LpJie?_GL}#LIxdL8;Jha{-4!L`otIo1zneWx_;SM;7R8 z36iwXq#Pp`!^Z4~N_uiviRCobi=J6(f*i>#_*cbhq^lGAcjLfLR705viIW<4tuQ0n zId6D~+yQ9rVRW{@loXhZS*w(1p#MqU#W8-I1yt|XAx7{i#>_-IXT+pEMYQ4YJ_F9w zA_TX1K9oLeMEZtL3qk~{ z&(o5D+KSmbbIHRQA;hJeO!8EHVRWh%Uc{Cr+tLgoDkKI*CHe1hiclN6ODY~WnINIF zJU#EoqWYv{eRh99qxkBm^hDv_H`I$5@A*NuiHCU9HXvr*kBoxPf-T`lK0j76D5cqx zF@iRA0lHgp5=@gd_bod`c2L2RNV%h^ct(hb3n};OLUdhn1Z%u099BD%M=72mP{%*6 zjt8vYe#GM5W$kJs`yPI&;-shWl3t$)F-e#POWj%m<)-( z3@RL@Z!UoS?NWvr6nc3LWTE(@-<&oIhbB$OWxqrnT;w3KzRHfIZJ3VTCu-bW6Y*)r z#C>H?e0TpiB-gcoxdmPC8@d1@XSxjlMkQ6+gYDsRsz!1@PaZBd&$1+!A4RepZdn@$ z)eZaJ5h_W7$&_3rLO8>8Mn97aSh-eC?ImEG4#m&2!V%E6=Z7L)rhnD+}|I(j^4cg0JS@3i2R(E0n#le~KMj~cpGFoLE zHylhB!ih;Os`BAnsi}#b)>&XN?x!TpVBBR|o85UK0&mJ~tV=pLSy}e70(2G%C^rhC zP>i*Xw>EOmz@WNzR<|2qoU0hVz8F`0B?)^S6YWZefP@-Pjw5fXl#XJXIANqP9ea)q zh^96wz?#kqw;tDueaEew_~SHc1DD^~(HW&nRa6y}o6%JqYBKPVZQdagqO?qn@hF{a z{^3ywCKaX8{PDsofDl7c{8D=e@jPDJz}l_ZY8KE&iB`83{|S81IqatJ`=Y zbc^#>ALIPU9C&nVsY;X-E~A?~Xd z77)L*!bY3PztA45f?WlN#cjMPa>h97&#LTo{wQk=6LbN{UFu6u&vaMaZ8}j<(IL6) z!B*iH?S}CUx1E-L15R*$*zU^995oGT;$b_YKQp6Iz_x!2U;)1)Hyq2MHjriHnQSlW zfB6U;+&Q?TW{F&T{2~7w#l{9kd$ihK1XchGF$POU0xGL-w(mdA&6&Z$C|19j^CSV91Z3rHNCSkGH&D zKVd^9lBJbG+7k-X>MTT3k>Z8jfh+=HD#HGg42OQrXXkcfW2urWUX9i9-=RuHx8}8F zoBTcJ!`8b6L-XmCzM{8m5)yUJ+Yx5~RO_*xKe8?p2AEI!y*)AQUHJGZB55M|e`H9m zsTZyRhnnKP3mb~Ya5@{S7SskWD>GtWeUlRmDp8a#ad9JjOOx%0A7np%2}*@z@jD~bBS z^3_D&hA)q_TgtHt!!^bzNLl*T$hMMQEq%$t4e?dUO_JplDREr{y&^iKOmr#y(0wpJ z5gt`pDv^Eh(@YOD`|}THILfd06SauEwRK-!*J2NJO!tY+XzP-f6~tQ1lNO;Mzq%&O zw*!^?1N>%;?FKK|*Hmi7MaRBy!lRCN$}7Xi@dfh<;rN64Us_sEQQjoZo=#&Q$c3`k z94gP;`iLa#<4O%ti3UH$^9B@Sp<6GXqb=4{oCNL&UxRGad|;OA%u$JU*`$%-^e+MP z?5b<@L7`vW%4R2F#9lBQ$1yiN)*|xVMNxCEaGzlYZzrFd_V`0yrO*3bW9%E{(7S}`(!^zn)K0HbG1qCzIp*j-Fpk#Hz&scM% z6P(gLjWv{v69*8ElI%asJQ4Ejc`WR22mN+~pxh;3| z2_<*Ag2$@@<@YE*m&AiBCC^HG4jZ&0u!B&;OFM%Qx;weca*=bji`f?85N*zZ$26By zK<{MmlDeUfXuF*to3lH*x+~?-h7u0Vhv$XGlSF?*JjCXq)rkz2MTj3Ma#~9*fomC1 z3bEEIk}3f%vl??e_Pv;}dAW2apanyTo=QR5`|NkJ`eAWdIoI7~#0lpPeSUSLZef(Q z#Lc;>-lT}b;~J59E2|`dA4t>{K0plm;((@(doG8hjyHJnvsZ~rX~=Za&!9S}#r+o2 zvzw``h(O+mhKoFvTT8kezd39K4hsEoGu&EEq}BPwA^iIE@nr8RbS7~OA8&gV8{gK( z$C`bm#xmKLJ)*)0+dW!Um;NS=h)28am=ST#_?R;c#dehk%Owi6XS2C;J`~_%#Dy7Z zGK~8}WyLmP$Xwjp<_!eN!8TFXZw$I9=Afj9eBlGH$VhQ}tz-;Ge6!%Hd-~X&@I^te z>J_W^j?xOJW`kpxow|y=})sQHei9s?n;b) z!6_&4Lb=%7=mvW!y^F8CylPJ>hgR?1vMZqRDLawF)gPOGEnhl>` zb@kPc0}~t3J!RIqE%`K9Bl4n*Ys4pJPpb}vC&okSYoc1M)yM|q5LhQAUA`Ev{H)qI zRq36JjN>QSzH(qgO`CUgQ2IzNncwP!$W32Gt+BsMZnqLv?v|~Kcq;Y-Yv7|*io3Un z$|9{!vduQ;CIOdZJ;0JZp{J@o+)D&v&^XNBgb5 zLGrivyJ^l9aWXF2p6G2)rR_W-=1zCf1?{A1}a$bBAQ(Y^(x zzE!4TK~A{4zc*gix>f)M*u#+~dB9!590h9VTr9_to6GKnMY3*80G)D9z;13C(H$=p zm2WV4iC5fjx$VkK!!4Q@rVt=x)5*ocPFL|+d+x`TbFl4?5F}f8`Se*V@F?}4DqLO- z4wd?z(M^=-lNU_!q>KHcNM@qZC=2M0sL^j@*BzdUdNMd z0>dhF6)X5q>(eI2;*D4KU8$y&T($;!=*c054$|E61m_(Qk$|7q6#{5JfiTKmI(O9A|s z%(tx+mSL;Q*#0D%6r*?<86?&QB+{XcE7 BbJ+j@ literal 0 HcmV?d00001 diff --git a/datasets/subjqa/subjqa.py b/datasets/subjqa/subjqa.py new file mode 100644 index 00000000000..642759c4046 --- /dev/null +++ b/datasets/subjqa/subjqa.py @@ -0,0 +1,211 @@ +# coding=utf-8 +# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +"""SubjQA is a question answering dataset that focuses on subjective questions and answers. +The dataset consists of roughly 10,000 questions over reviews from 6 different domains: books, movies, grocery, +electronics, TripAdvisor (i.e. hotels), and restaurants.""" + + +import ast +import os + +import pandas as pd + +import datasets + + +_CITATION = """\ +@inproceedings{bjerva20subjqa, + title = "SubjQA: A Dataset for Subjectivity and Review Comprehension", + author = "Bjerva, Johannes and + Bhutani, Nikita and + Golahn, Behzad and + Tan, Wang-Chiew and + Augenstein, Isabelle", + booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing", + month = November, + year = "2020", + publisher = "Association for Computational Linguistics", +} +""" + +_DESCRIPTION = """SubjQA is a question answering dataset that focuses on subjective questions and answers. +The dataset consists of roughly 10,000 questions over reviews from 6 different domains: books, movies, grocery, +electronics, TripAdvisor (i.e. hotels), and restaurants.""" + +_HOMEPAGE = "" + +_LICENSE = "" + +_URLs = {"default": "https://github.com/lewtun/SubjQA/archive/refs/heads/master.zip"} + + +class Subjqa(datasets.GeneratorBasedBuilder): + """SubjQA is a question answering dataset that focuses on subjective questions and answers.""" + + VERSION = datasets.Version("1.1.0") + + BUILDER_CONFIGS = [ + datasets.BuilderConfig(name="books", version=VERSION, description="Amazon book reviews"), + datasets.BuilderConfig(name="electronics", version=VERSION, description="Amazon electronics reviews"), + datasets.BuilderConfig(name="grocery", version=VERSION, description="Amazon grocery reviews"), + datasets.BuilderConfig(name="movies", version=VERSION, description="Amazon movie reviews"), + datasets.BuilderConfig(name="restaurants", version=VERSION, description="Yelp restaurant reviews"), + datasets.BuilderConfig(name="tripadvisor", version=VERSION, description="TripAdvisor hotel reviews"), + ] + + def _info(self): + features = datasets.Features( + { + "domain": datasets.Value("string"), + "nn_mod": datasets.Value("string"), + "nn_asp": datasets.Value("string"), + "query_mod": datasets.Value("string"), + "query_asp": datasets.Value("string"), + "q_reviews_id": datasets.Value("string"), + "question_subj_level": datasets.Value("int64"), + "ques_subj_score": datasets.Value("float"), + "is_ques_subjective": datasets.Value("bool"), + "review_id": datasets.Value("string"), + "id": datasets.Value("string"), + "title": datasets.Value("string"), + "context": datasets.Value("string"), + "question": datasets.Value("string"), + "answers": datasets.features.Sequence( + { + "text": datasets.Value("string"), + "answer_start": datasets.Value("int32"), + "answer_subj_level": datasets.Value("int64"), + "ans_subj_score": datasets.Value("float"), + "is_ans_subjective": datasets.Value("bool"), + } + ), + } + ) + return datasets.DatasetInfo( + description=_DESCRIPTION, + features=features, + supervised_keys=None, + homepage=_HOMEPAGE, + license=_LICENSE, + citation=_CITATION, + ) + + def _split_generators(self, dl_manager): + data_dir = dl_manager.download_and_extract(_URLs["default"]) + return [ + datasets.SplitGenerator( + name=datasets.Split.TRAIN, + gen_kwargs={ + "filepath": os.path.join(data_dir, f"SubjQA-master/SubjQA/{self.config.name}/splits/train.csv") + }, + ), + datasets.SplitGenerator( + name=datasets.Split.TEST, + gen_kwargs={ + "filepath": os.path.join(data_dir, f"SubjQA-master/SubjQA/{self.config.name}/splits/test.csv") + }, + ), + datasets.SplitGenerator( + name=datasets.Split.VALIDATION, + gen_kwargs={ + "filepath": os.path.join(data_dir, f"SubjQA-master/SubjQA/{self.config.name}/splits/dev.csv") + }, + ), + ] + + def _generate_examples(self, filepath): + df = pd.read_csv(filepath) + squad_format = self._convert_to_squad(df) + for example in squad_format["data"]: + title = example.get("title", "").strip() + for paragraph in example["paragraphs"]: + context = paragraph["context"].strip() + for qa in paragraph["qas"]: + question = qa["question"].strip() + question_meta = {k: v for k, v in qa.items() if k in self.question_meta_columns} + id_ = qa["id"] + answer_starts = [answer["answer_start"] for answer in qa["answers"]] + answers = [answer["text"].strip() for answer in qa["answers"]] + answer_meta = pd.DataFrame(qa["answers"], columns=self.answer_meta_columns).to_dict("list") + yield id_, { + **{ + "title": title, + "context": context, + "question": question, + "id": id_, + "answers": { + **{ + "answer_start": answer_starts, + "text": answers, + }, + **answer_meta, + }, + }, + **question_meta, + } + + def _create_paragraphs(self, df): + "A helper function to convert a pandas.DataFrame of (question, context, answer) rows to SQuAD paragraphs." + self.question_meta_columns = [ + "domain", + "nn_mod", + "nn_asp", + "query_mod", + "query_asp", + "q_reviews_id", + "question_subj_level", + "ques_subj_score", + "is_ques_subjective", + "review_id", + ] + self.answer_meta_columns = ["answer_subj_level", "ans_subj_score", "is_ans_subjective"] + id2review = dict(zip(df["review_id"], df["review"])) + pars = [] + for review_id, review in id2review.items(): + qas = [] + review_df = df.query(f"review_id == '{review_id}'") + id2question = dict(zip(review_df["q_review_id"], review_df["question"])) + + for k, v in id2question.items(): + d = df.query(f"q_review_id == '{k}'").to_dict(orient="list") + answer_starts = [ast.literal_eval(a)[0] for a in d["human_ans_indices"]] + answer_meta = {k: v[0] for k, v in d.items() if k in self.answer_meta_columns} + question_meta = {k: v[0] for k, v in d.items() if k in self.question_meta_columns} + # Only fill answerable questions + if pd.unique(d["human_ans_spans"])[0] != "ANSWERNOTFOUND": + answers = [ + {**{"text": text, "answer_start": answer_start}, **answer_meta} + for text, answer_start in zip(d["human_ans_spans"], answer_starts) + if text != "ANSWERNOTFOUND" + ] + else: + answers = [] + qas.append({**{"question": v, "id": k, "answers": answers}, **question_meta}) + # Slice off ANSWERNOTFOUND from context + pars.append({"qas": qas, "context": review[: -len(" ANSWERNOTFOUND")]}) + return pars + + def _convert_to_squad(self, df): + "A helper function to convert a pandas.DataFrame of product-based QA dataset into SQuAD format" + groups = ( + df.groupby("item_id") + .apply(self._create_paragraphs) + .to_frame(name="paragraphs") + .reset_index() + .rename(columns={"item_id": "title"}) + ) + squad_data = {} + squad_data["data"] = groups.to_dict(orient="records") + return squad_data