--- license: other dataset_info: - config_name: 2WikiMultihopQA features: - name: _id dtype: string - name: type dtype: string - name: question dtype: string - name: context sequence: - name: title dtype: string - name: content sequence: string - name: supporting_facts sequence: - name: title dtype: string - name: sent_id dtype: int32 - name: evidences sequence: - name: fact dtype: string - name: relation dtype: string - name: entity dtype: string - name: answer dtype: string splits: - name: train num_bytes: 662142981 num_examples: 167454 - name: dev num_bytes: 54346346 num_examples: 12576 - name: test num_bytes: 51639331 num_examples: 12576 download_size: 389826062 dataset_size: 768128658 - config_name: MuSiQue features: - name: id dtype: string - name: paragraphs list: - name: idx dtype: int64 - name: title dtype: string - name: paragraph_text dtype: string - name: is_supporting dtype: bool - name: question dtype: string - name: question_decomposition list: - name: id dtype: int64 - name: question dtype: string - name: answer dtype: string - name: paragraph_support_idx dtype: int64 - name: answer dtype: string - name: answer_aliases sequence: string - name: answerable dtype: bool - name: text_all dtype: string - name: text_all_support dtype: string splits: - name: validation num_bytes: 55971326 num_examples: 2417 download_size: 23776203 dataset_size: 55971326 - config_name: NQ features: - name: id dtype: string - name: title dtype: string - name: document dtype: string - name: question dtype: string - name: long_answers sequence: string - name: short_answers sequence: string - name: retrieved_passages sequence: string splits: - name: validation num_bytes: 279214996 num_examples: 4289 download_size: 141438208 dataset_size: 279214996 - config_name: hotpotqa features: - name: id dtype: string - name: question dtype: string - name: answer dtype: string - name: type dtype: string - name: level dtype: string - name: supporting_facts sequence: - name: title dtype: string - name: sent_id dtype: int32 - name: context sequence: - name: title dtype: string - name: sentences sequence: string - name: rag sequence: string - name: retrieved_passages sequence: string splits: - name: validation num_bytes: 131225660 num_examples: 7405 download_size: 77113296 dataset_size: 131225660 - config_name: triviaqa features: - name: question dtype: string - name: question_id dtype: string - name: question_source dtype: string - name: entity_pages sequence: - name: doc_source dtype: string - name: filename dtype: string - name: title dtype: string - name: wiki_context dtype: string - name: search_results sequence: - name: description dtype: string - name: filename dtype: string - name: rank dtype: int32 - name: title dtype: string - name: url dtype: string - name: search_context dtype: string - name: answer struct: - name: aliases sequence: string - name: normalized_aliases sequence: string - name: matched_wiki_entity_name dtype: string - name: normalized_matched_wiki_entity_name dtype: string - name: normalized_value dtype: string - name: type dtype: string - name: value dtype: string - name: retrieved_passages sequence: string splits: - name: validation num_bytes: 474767227 num_examples: 7993 download_size: 262352984 dataset_size: 474767227 - config_name: truthfulqa features: - name: question dtype: string - name: mc1_targets struct: - name: choices sequence: string - name: labels sequence: int32 - name: mc2_targets struct: - name: choices sequence: string - name: labels sequence: int32 - name: category dtype: string - name: source dtype: string - name: website_data dtype: string - name: retrieved_passages sequence: string splits: - name: validation num_bytes: 24476993 num_examples: 817 download_size: 10176147 dataset_size: 24476993 configs: - config_name: 2WikiMultihopQA data_files: - split: train path: 2WikiMultihopQA/train-* - split: dev path: 2WikiMultihopQA/dev-* - split: test path: 2WikiMultihopQA/test-* - config_name: MuSiQue data_files: - split: validation path: MuSiQue/validation-* - config_name: NQ data_files: - split: validation path: NQ/validation-* - config_name: boolq data_files: - split: validation path: boolq/validation-* - config_name: hotpotqa data_files: - split: validation path: hotpotqa/validation-* - config_name: triviaqa data_files: - split: validation path: triviaqa/validation-* - config_name: truthfulqa data_files: - split: validation path: truthfulqa/validation-* --- # ContextualBench - A comprehensive toolkit to evaluate LM on different Contextual datasets Evaluation Code: [SalesforceAIResearch/SFR-RAG](https://github.com/SalesforceAIResearch/SFR-RAG) ## Description ContextualBench is a powerful evaluation framework designed to assess the performance of Large Language Models (LLMs) on contextual datasets. It provides a flexible pipeline for evaluating various LLM families across different tasks, with a focus on handling large context inputs. > Users need to make their own assessment regarding any obligations or responsibilities under the corresponding licenses or terms and conditions pertaining to the original datasets and data. ## Features * Dynamic Retrieval Support: Efficiently handles large context inputs, allowing for comprehensive evaluation of LLMs' contextual understanding capabilities. * Extensive Evaluation Dataset: Supports 7 contextual tasks, including: Question Answering (QA), Multi-Hop Question Answering, Classification tasks * Multi-LLM Family Support: Compatible with a wide range of LLM families, including: Hugging Face models, Gemma, Mistral, OpenAI, Cohere. The dataset can be loaded using the command ```python task = "hotpotqa" # it can be any other option load_dataset("Salesforce/ContextualBench", task, split="validation") ``` ## Component Datasets of ContextualBench > Each individual evaluation dataset in ContextualBench is licensed separately and must be adhered by a user. ### 2WikiHotpotQA This dataset is a multihop question answering task, as proposed in "Constructing A Multi-hop QA Dataset for Comprehensive Evaluation of Reasoning Steps" by Ho. et. al The folder contains evaluation script and path to dataset on the validation split on around 12k samples. ``` @inproceedings{xanh2020_2wikimultihop, title = "Constructing A Multi-hop {QA} Dataset for Comprehensive Evaluation of Reasoning Steps", author = "Ho, Xanh and Duong Nguyen, Anh-Khoa and Sugawara, Saku and Aizawa, Akiko", booktitle = "Proceedings of the 28th International Conference on Computational Linguistics", month = dec, year = "2020", address = "Barcelona, Spain (Online)", publisher = "International Committee on Computational Linguistics", url = "https://www.aclweb.org/anthology/2020.coling-main.580", pages = "6609--6625", } ``` ### HotpotQA HotpotQA is a Wikipedia-based question-answer pairs with the questions require finding and reasoning over multiple supporting documents to answer. We evaluate on 7405 datapoints, on the distractor setting. This dataset was proposed in the below paper ``` @inproceedings{yang2018hotpotqa, title={{HotpotQA}: A Dataset for Diverse, Explainable Multi-hop Question Answering}, author={Yang, Zhilin and Qi, Peng and Zhang, Saizheng and Bengio, Yoshua and Cohen, William W. and Salakhutdinov, Ruslan and Manning, Christopher D.}, booktitle={Conference on Empirical Methods in Natural Language Processing ({EMNLP})}, year={2018} } ``` ### MuSiQue This dataset is a multihop question answering task, that requires 2-4 hop in every questions, making it slightly harder task when compared to other multihop tasks.This dataset was proposed in the below paper ``` @article{trivedi2021musique, title={{M}u{S}i{Q}ue: Multihop Questions via Single-hop Question Composition}, author={Trivedi, Harsh and Balasubramanian, Niranjan and Khot, Tushar and Sabharwal, Ashish}, journal={Transactions of the Association for Computational Linguistics}, year={2022} publisher={MIT Press} } ``` ### NaturalQuestions The NQ corpus contains questions from real users, and it requires QA systems to read and comprehend an entire Wikipedia article that may or may not contain the answer to the question ``` @article{47761, title = {Natural Questions: a Benchmark for Question Answering Research}, author = {Tom Kwiatkowski and Jennimaria Palomaki and Olivia Redfield and Michael Collins and Ankur Parikh and Chris Alberti and Danielle Epstein and Illia Polosukhin and Matthew Kelcey and Jacob Devlin and Kenton Lee and Kristina N. Toutanova and Llion Jones and Ming-Wei Chang and Andrew Dai and Jakob Uszkoreit and Quoc Le and Slav Petrov}, year = {2019}, journal = {Transactions of the Association of Computational Linguistics} } ``` ### PopQA PopQA is a large-scale open-domain question answering (QA) dataset, the long-tail subset, consisting of 1,399 rare entity queries whose monthly Wikipedia page views are less than 100 Make sure to cite the work ``` @article{ mallen2023llm_memorization , title={When Not to Trust Language Models: Investigating Effectiveness and Limitations of Parametric and Non-Parametric Memories }, author={ Mallen, Alex and Asai,Akari and Zhong, Victor and Das, Rajarshi and Hajishirzi, Hannaneh and Khashabi, Daniel}, journal={ arXiv preprint }, year={ 2022 } } ``` ### TriviaQA TriviaqQA is a reading comprehension dataset containing question-answer pairs authored by trivia enthusiasts and independently gathered evidence documents, six per question on average, that provide high quality distant supervision for answering the questions. ``` @article{2017arXivtriviaqa, author = {{Joshi}, Mandar and {Choi}, Eunsol and {Weld}, Daniel and {Zettlemoyer}, Luke}, title = "{triviaqa: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension}", journal = {arXiv e-prints}, year = 2017, eid = {arXiv:1705.03551}, pages = {arXiv:1705.03551}, archivePrefix = {arXiv}, eprint = {1705.03551}, } ``` ### TruthfulQA TruthfulQA is a benchmark to measure whether a language model is truthful in generating answers to questions. The benchmark comprises 817 questions that span 38 categories, including health, law, finance and politics. Questions are crafted so that some humans would answer falsely due to a false belief or misconception. To perform well, models must avoid generating false answers learned from imitating human texts. ``` @misc{lin2021truthfulqa, title={TruthfulQA: Measuring How Models Mimic Human Falsehoods}, author={Stephanie Lin and Jacob Hilton and Owain Evans}, year={2021}, eprint={2109.07958}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` ## Citation ``` @article{nguyen2024sfrrag, title={SFR-RAG: Towards Contextually Faithful LLMs}, author={Nguyen, Xuan-Phi and Pandit, Shrey and Purushwalkam, Senthil and Xu, Austin and Chen, Hailin and Ming, Yifei and Ke, Zixuan and Savarese, Silvio and Xong, Caiming and Joty, Shafiq}, year={2024} } ```