search

guotong1988 / NL2SQL-RULE

Content Enhanced BERT-based Text-to-SQL Generation https://arxiv.org/abs/1910.07179
188 stars 48 forks source link

missing training data #3

Closed beyhangl closed 4 years ago

beyhangl commented 4 years ago

Hey,

Could verify the dataset

'./data_and_model/train_tok.jsonl'

'./data_and_model/train_knowledge.jsonl'

These are same ? Because your code reading train_tok but it's missing

guotong1988 commented 4 years ago

No. train_knowledge.jsonl' is generated fromtrain_tok.jsonl. You can get train_tok.jsonl from https://drive.google.com/file/d/1iJvsf38f16el58H4NPINQ7uzal5-V4v4

beyhangl commented 4 years ago

thanks for quick response. I just wondering about how can we test it with our own data dictionary and example sentence?

guotong1988 commented 4 years ago

https://github.com/guotong1988/NL2SQL-BERT/blob/master/data_and_model/output_entity.py

You can only cache the more important data.

fromfuture-ai commented 4 years ago

Has anyone managed to run this code, can please help?

I have cloned the code, downloaded the file from gDrive, put it in the folder, downloaded Bert another folder.... and then run the python3 train.py ..... command. It ran, for a few seconds. And does not create any output.

Can the author please have a look at the readme, giving a fresh look. I failed to follow. I have downloaded missing datasets, etc. and still not working. Sorry, a bit frustrating.