defog/sqlcoder2

1---2license: other3language:4- en5pipeline_tag: text-generation6tags:7- code8---9# Defog SQLCoder10Defog's SQLCoder is a state-of-the-art LLM for converting natural language questions to SQL queries.11 12[Interactive Demo](https://defog.ai/sqlcoder-demo/) | [🤗 HF Repo](https://huggingface.co/defog/sqlcoder2) | [♾️ Colab](https://colab.research.google.com/drive/1z4rmOEiFkxkMiecAWeTUlPl0OmKgfEu7?usp=sharing) | [🐦 Twitter](https://twitter.com/defogdata)13 14## TL;DR15SQLCoder is a 15B parameter model that outperforms `gpt-3.5-turbo` for natural language to SQL generation tasks on our [sql-eval](https://github.com/defog-ai/sql-eval) framework, and significantly outperforms all popular open-source models. When fine-tuned on a given schema, it also outperforms `gpt-4`16 17SQLCoder is fine-tuned on a base StarCoder model.18 19## Results on novel datasets not seen in training20| model   | perc_correct |21|-|-|  22| gpt4-2023-10-04    | 82.0 |23| defog-sqlcoder2    | 77.5 |24| gpt4-2023-08-28    | 74.0 |25| defog-sqlcoder-7b  | 71.0 |26| gpt-3.5-2023-10-04 | 66.0 |27| claude-2           | 64.5 |28| gpt-3.5-2023-08-28 | 61.0 |29| claude_instant_1   | 61.0 |30| text-davinci-003   | 52.5 |31 32## License33The 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, with additional responsible use restrictions added. 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.34 35## Training36Defog 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. 37 38You can read more about our [training approach](https://defog.ai/blog/open-sourcing-sqlcoder2-7b/) and [evaluation framework](https://defog.ai/blog/open-sourcing-sqleval/).39 40## Results by question category41We classified each generated question into one of 5 categories. The table displays the percentage of questions answered correctly by each model, broken down by category.42| query_category   |   gpt-4 |   sqlcoder2-15b |   sqlcoder-7b |   gpt-3.5 |   claude-2 |   claude-instant |   gpt-3 |43|:-----------------|--------:|----------------:|--------------:|----------:|-----------:|-----------------:|--------:|44| date             |    72   |            80   |          64   |      68   |       52   |             48   |    32   |45| group_by         |    91.4 |            82.9 |          82.9 |      77.1 |       71.4 |             71.4 |    71.4 |46| order_by         |    82.9 |            77.1 |          74.3 |      68.6 |       74.3 |             74.3 |    68.6 |47| ratio            |    80   |            74.3 |          54.3 |      37.1 |       57.1 |             45.7 |    25.7 |48| join             |    82.9 |            74.3 |          74.3 |      71.4 |       65.7 |             62.9 |    57.1 |49| where            |    80   |            77.1 |          74.3 |      74.3 |       62.9 |             60   |    54.3 |50 51## Using SQLCoder52You can use SQLCoder via the `transformers` library by downloading our model weights from the Hugging Face repo. We have added sample code for [inference](https://github.com/defog-ai/sqlcoder/blob/main/inference.py) on a [sample database schema](https://github.com/defog-ai/sqlcoder/blob/main/metadata.sql). 53```bash54python inference.py -q "Question about the sample database goes here"55 56# Sample question:57# 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.58```59 60You can also use a demo on our website [here](https://defog.ai/sqlcoder-demo), or run SQLCoder in Colab [here](https://colab.research.google.com/drive/13BIKsqHnPOBcQ-ba2p77L5saiepTIwu0#scrollTo=ZpbVgVHMkJvC)61 62## Hardware Requirements63SQLCoder has been tested on an A100 40GB GPU with `bfloat16` 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.64 65## Todo66 67- [x] Open-source the v1 model weights68- [x] Train the model on more data, with higher data variance69- [ ] Tune the model further with Reward Modelling and RLHF70- [ ] Pretrain a model from scratch that specializes in SQL analysis