--- annotations_creators: - no-annotation language_creators: - found language: - en license: - apache-2.0 multilinguality: - monolingual pretty_name: Fewshot Table Dataset size_categories: - 100K 20% non-English text as measured by [SpaCy](https://spacy.io/) c) Given 2 Million passing tables we consider each table column as a potential output column, and concatenate all other columns to form the input (which produces 5.6 M candidate tasks) 5. Rule-based-checks to reject tasks a) We reject a task if it has less than 6 rows. Note that tasks may have fewer rows than their origin tables since we remove rows where the output column is empty. b) We reject tasks if any input maps to multiple outputs. c) We reject tasks if it has fewer than 2 output classes. d) We reject a task if the output column alone has >20% non-English text. e) We reject a task if the classes are heavily imbalanced. 6. Lastly we apply domain-level filtering. Initial iterations of our dataset found a significant imbalance in terms of the website of origin for our generated tasks. In particular, we found that the mos-frequent domain in the WDC corpus, Cappex.com, was emphasized by our export criteria such that this website alone represented 41% of our total tasks. Since we want our dataset to represent the diversity of all the tables available on the web, we apply a hard fix for this imbalance by limiting the number of tasks per domain. Starting from the initial corpus of 50M tables from 323160 web domains, our resulting longlist of tasks comprises more than X for a total of 413350 tasks. #### Who are the source language producers? The dataset is extracted from [WDC Web Table Corpora](http://webdatacommons.org/webtables/). ### Annotations #### Annotation process No annotation Process #### Who are the annotators? - ### Personal and Sensitive Information The data was extracted from [WDC Web Table Corpora](http://webdatacommons.org/webtables/), which in turn extracted tables from the [Common Crawl](https://commoncrawl.org/). We did not filter the data in any way. Thus any user identities or otherwise sensitive information (e.g. data that reveals racial or ethnic origins, sexual orientations, religious beliefs, political opinions or union memberships, or locations; financial or health data; biometric or genetic data; forms of government identification, such as social security numbers; criminal history, etc.) might be contained in our dataset. ## Considerations for Using the Data ### Social Impact of Dataset The purpose of this dataset is to help develop models that are better at few-shot learning and have higher few-shot performance by fine-tuning few-shot tasks extracted from tables. While tables have a similar structure to few-shot tasks and we do see an improved performance on few-shot tasks in our paper, we want to make clear that finetuning on tables also has its risks. First of all, since the tables are extracted from the web, they may contain user identities or otherwise sensitive information which a model might reveal at inference, or which could influence the learning process of a model in a negative way. Second, since tables are very diverse in nature, the model also trains on low-quality data or data with an unusual structure. While it is interesting that training on such data improves few-shot performance on downstream tasks, this could also imply that the model learns concepts that are very dissimilar to human concepts that would be useful for a certain downstream task. In other words, it is possible that the model learns weird things that are helpful on the evaluated downstream tasks, but might lead to bad out-of-distribution behavior. ### Discussion of Biases Since our dataset contains tables that are scraped from the web, it will also contain many toxic, racist, sexist, and otherwise harmful biases and texts. We have not run any analysis on the biases prevalent in our datasets. Neither have we explicitly filtered the content for toxic content. This implies that a model trained on our dataset will reinforce harmful biases and toxic text that exist in our dataset. ### Other Known Limitations [Needs More Information] ## Additional Information ### Dataset Curators Mention all authors ### Licensing Information Apache 2.0 ### Citation Information [Needs More Information]