TAPAS base model

This model has 2 versions which can be used. The latest version, which is the default one, corresponds to the tapas_inter_masklm_base_reset checkpoint of the original Github repository . This model was pre-trained on MLM and an additional step which the authors call intermediate pre-training. It uses relative position embeddings by default (i.e. resetting the position index at every cell of the table).

The other (non-default) version which can be used is the one with absolute position embeddings:

• revision="no_reset" , which corresponds to tapas_inter_masklm_base

Disclaimer: The team releasing TAPAS did not write a model card for this model so this model card has been written by the Hugging Face team and contributors.

Model description

TAPAS is a BERT-like transformers model pretrained on a large corpus of English data from Wikipedia in a self-supervised fashion. This means it was pretrained on the raw tables and associated texts only, with no humans labelling them in any way (which is why it can use lots of publicly available data) with an automatic process to generate inputs and labels from those texts. More precisely, it was pretrained with two objectives:

• Masked language modeling (MLM): taking a (flattened) table and associated context, the model randomly masks 15% of the words in the input, then runs the entire (partially masked) sequence through the model. The model then has to predict the masked words. This is different from traditional recurrent neural networks (RNNs) that usually see the words one after the other, or from autoregressive models like GPT which internally mask the future tokens. It allows the model to learn a bidirectional representation of a table and associated text.
• Intermediate pre-training: to encourage numerical reasoning on tables, the authors additionally pre-trained the model by creating a balanced dataset of millions of syntactically created training examples. Here, the model must predict (classify) whether a sentence is supported or refuted by the contents of a table. The training examples are created based on synthetic as well as counterfactual statements.

This way, the model learns an inner representation of the English language used in tables and associated texts, which can then be used to extract features useful for downstream tasks such as answering questions about a table, or determining whether a sentence is entailed or refuted by the contents of a table. Fine-tuning is done by adding one or more classification heads on top of the pre-trained model, and then jointly train these randomly initialized classification heads with the base model on a downstream task.

Intended uses & limitations

You can use the raw model for getting hidden representatons about table-question pairs, but it's mostly intended to be fine-tuned on a downstream task such as question answering or sequence classification. See the model hub to look for fine-tuned versions on a task that interests you.

Training procedure

Preprocessing

The texts are lowercased and tokenized using WordPiece and a vocabulary size of 30,000. The inputs of the model are then of the form:

[CLS] Sentence [SEP] Flattened table [SEP]

Pre-training

The model was pre-trained on 32 Cloud TPU v3 cores for 1,000,000 steps with maximum sequence length 512 and batch size of 512. In this setup, pre-training on MLM only takes around 3 days. Aditionally, the model has been further pre-trained on a second task (table entailment). See the original TAPAS paper and the follow-up paper for more details.

The optimizer used is Adam with a learning rate of 5e-5, and a warmup ratio of 0.01.

BibTeX entry and citation info

@misc{herzig2020tapas,
      title={TAPAS: Weakly Supervised Table Parsing via Pre-training}, 
      author={Jonathan Herzig and Paweł Krzysztof Nowak and Thomas Müller and Francesco Piccinno and Julian Martin Eisenschlos},
      year={2020},
      eprint={2004.02349},
      archivePrefix={arXiv},
      primaryClass={cs.IR}
}

@misc{eisenschlos2020understanding,
      title={Understanding tables with intermediate pre-training}, 
      author={Julian Martin Eisenschlos and Syrine Krichene and Thomas Müller},
      year={2020},
      eprint={2010.00571},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}