Skip to content

The code for SIGMOD 2024 paper titled "Watchog: A Light-weight Contrastive Learning based Framework for Column Annotation"

Notifications You must be signed in to change notification settings

megagonlabs/watchog

Repository files navigation

watchog

This is the repo for the code of Watchog: A Light-weight Contrastive Learning based Framework for Table Understanding.

Run fine-tuning with contrastive learning

See run_supcl_train.py

Run fine-tuning for down-stream tasks

See run_finetune.py for semantic type detection and relationship extraction on full datasets (SATO and TURL). See run_finetune_small.py for sampled datasets and run_finetune_ssl.py for sampled datasets under semi-supervised setting.

About

The code for SIGMOD 2024 paper titled "Watchog: A Light-weight Contrastive Learning based Framework for Column Annotation"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages