This repository contains source code for the TaBERT model, a pre-trained language model for learning joint representations of natural language utterances and (semi-)structured tables for semantic parsing. TaBERT is pre-trained on a massive corpus of 26M Web tables and their associated natural language context, and could be used as a drop-in replacement of a semantic parsers original encoder to compute representations for utterances and table schemas (columns).
You said you use n-gram overlap ratio with the utterance to select the top-K rows for K > 1 in the paper, but after inspecting the code, I found that you only choose the first k row in encoding a table , so I'm a little confused.
I'd appreciate it if you can tell more details of content snapshot generating.
You said you use n-gram overlap ratio with the utterance to select the top-K rows for K > 1 in the paper, but after inspecting the code, I found that you only choose the first k row in encoding a table , so I'm a little confused. I'd appreciate it if you can tell more details of content snapshot generating.