Transformers: Text-to-SQL Query

Created on 7 Aug 2020  ยท  9Comments  ยท  Source: huggingface/transformers

โ“ Questions & Help

Hello everyone, I got a task that I want to use NLP for convert text into a SQL query. Anyone knows how to do this or got any suggestion? Thanks.

wontfix

Most helpful comment

hi @thiagomoeng, can you try setting the xla_device argument in config.json to False.

```python3
config = T5Config.from_pretrained("mrm8488/t5-base-finetuned-wikiSQL")
config.xla_device = False
model = T5ForConditionalGeneration.from_pretrained("mrm8488/t5-base-finetuned-wikiSQL", config=config)

All 9 comments

pinging @mrm8488 .

You can try this model: https://huggingface.co/mrm8488/t5-base-finetuned-wikiSQL
Thanks for ping me @patil-suraj

If you go to the model hub and type 'SQL', you get this. There are currently 3 variants of the T5 model fine-tuned on WikiSQL (a large dataset that contains sentence - SQL pairs).

Hi @mrm8488 I am having some issue with "torch_xla" module, is there any way to run this model on local windows, without this TPU module? Thanks.

hi @thiagomoeng, can you try setting the xla_device argument in config.json to False.

```python3
config = T5Config.from_pretrained("mrm8488/t5-base-finetuned-wikiSQL")
config.xla_device = False
model = T5ForConditionalGeneration.from_pretrained("mrm8488/t5-base-finetuned-wikiSQL", config=config)

Ofc, I can but the basic structure band the most of the code is from the public colab about fine tuning T5 on TPU by @patil-suraj

Hi @mrm8488 I got a SQL data on Oracle, do you can give me some way of how to prepare this sql data to finetune on your model? I am beginner on training models.

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

@thiagomoeng if you want to convert natural language to sql , here is one implementation : https://github.com/abhijithneilabraham/tableQA

Drop me any comments on the slack channel in the readme there.

Was this page helpful?
0 / 5 - 0 ratings

Related issues

alphanlp picture alphanlp  ยท  3Comments

adigoryl picture adigoryl  ยท  3Comments

guanlongtianzi picture guanlongtianzi  ยท  3Comments

ereday picture ereday  ยท  3Comments

siddsach picture siddsach  ยท  3Comments