search

tshu-w / DBCopilot

Code and data for the paper "DBCᴏᴘɪʟᴏᴛ: Scaling Natural Language Querying to Massive Databases"
38 stars 6 forks source link

files not found for training data synthesis #3

Open Tsing-gu opened 3 months ago

Tsing-gu commented 3 months ago

It's failed to find checkpoint file when i try to train data synthesis.

Detail is as followed:

b77751ad14c3e1fc7f77defdeaa75f39

tshu-w commented 3 months ago

You should replace the checkpoint path of schema questioning models with your own path. https://github.com/tshu-w/DBCopilot/blob/2f65a4c71267eacac2cf5eddb17f0f37223e4dd4/scripts/synthesize_data.py#L110-L112

I will improve this script to accept command line arguments to customize these options.

wasedawesley commented 1 month ago

Hi, Where is the codes for training data systhesis?

tshu-w commented 1 month ago

Hi, Where is the codes for training data systhesis?

The code for model, data and config can be found it https://github.com/tshu-w/DBCopilot/blob/master/src/models/schema_questioning.py, https://github.com/tshu-w/DBCopilot/blob/master/src/datamodules/schema2text.py, and https://github.com/tshu-w/DBCopilot/blob/master/configs/schema_questioning.yaml, respectively.