Defog SQLCoder

Updated on Nov 14 to reflect benchmarks for SQLCoder-34B

Defog's SQLCoder is a state-of-the-art LLM for converting natural language questions to SQL queries.

TL;DR

SQLCoder is a family of large language models that outperforms gpt-4 and gpt-4-turbo for natural language to SQL generation tasks on our sql-eval framework, and significantly outperform all popular open-source models.

SQLCoder-70B and SQLCoder-34B are fine-tuned on a base CodeLlama model.

Results on novel datasets not seen in training

model	perc_correct
defog-sqlcoder-70b	93.0
defog-sqlcoder-34b	84.0
gpt4-2024-01-30	82.0
gpt4-turbo-2024-01-30	78.0
defog-sqlcoder2	77.5
defog-sqlcoder-7b	71.0
gpt-3.5-2024-01-30	65.0
claude-2	64.5
gpt-3.5-2023-08-28	61.0
claude_instant_1	61.0
text-davinci-003	52.5

]

License

The code in this repo (what little there is of it) is Apache-2 licensed. The model weights have a CC BY-SA 4.0 license. The TL;DR is that you can use and modify the model for any purpose – including commercial use. However, if you modify the weights (for example, by fine-tuning), you must open-source your modified weights under the same license terms.

Training

Defog was trained on more than 20,000 human-curated questions. These questions were based on 10 different schemas. None of the schemas in the training data were included in our evaluation framework.

You can read more about our training approach and evaluation framework.

Results by question category

We classified each generated question into one of 6 categories. The table displays the percentage of questions answered correctly by each model, broken down by category.

	date	group_by	order_by	ratio	join	where
sqlcoder-70b	96	91.4	97.1	85.7	97.1	91.4
sqlcoder-34b	80	94.3	85.7	77.1	85.7	80
gpt-4	64	94.3	88.6	74.2	85.7	80
sqlcoder2-15b	76	80	77.1	60	77.1	77.1
sqlcoder-7b	64	82.9	74.3	54.3	74.3	74.3
gpt-3.5	68	77.1	74.2	34.3	65.7	71.4
claude-2	52	71.4	74.3	57.1	65.7	62.9

Using SQLCoder

You can use SQLCoder via the transformers library by downloading our model weights from the Hugging Face repo. We have added sample code for inference on a sample database schema.

python inference.py -q "Question about the sample database goes here"

# Sample question:
# Do we get more revenue from customers in New York compared to customers in San Francisco? Give me the total revenue for each city, and the difference between the two.

You can also use a demo on our website here

Hardware Requirements

SQLCoder-34B has been tested on a 4xA10 GPU with float16 weights. You can also load an 8-bit and 4-bit quantized version of the model on consumer GPUs with 20GB or more of memory – like RTX 4090, RTX 3090, and Apple M2 Pro, M2 Max, or M2 Ultra Chips with 20GB or more of memory.

Todo

Open-source the v1 model weights
Train the model on more data, with higher data variance
Tune the model further with Reward Modelling and RLHF
Pretrain a model from scratch that specializes in SQL analysis