# Hugging Face — Efficient Table Pre-training without Real Data: An Introduction to TAPEX

- Company: Hugging Face (huggingface.co)
- Announced: 2022-05-23
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/tapex
- Record: https://forck.live/items/2193-efficient-table-pre-training-without-real-data-an-introduction-to-tapex
- Subject: Platform
- Models affected: TAPEX, BART

TAPEX, a table pre-training method that learns a neural SQL executor using synthetic SQL queries and execution results, achieving state-of-the-art results on table question answering and fact verification benchmarks.

## Evidence

Verbatim from https://huggingface.co/blog/tapex:

> In "TAPEX: Table Pre-training via Learning a Neural SQL Executor", we explore using synthetic data as a proxy for real data during pre-training, and demonstrate its powerfulness with TAPEX (Table Pre-training via Execution) as an example.

---

Record: https://forck.live/items/2193-efficient-table-pre-training-without-real-data-an-introduction-to-tapex
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md