File size: 4,504 Bytes
0440e12 d6b8247 a88aee5 d6b8247 7e2fb51 d6b8247 a88aee5 d6b8247 a88aee5 d6b8247 4ccba91 d6b8247 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 |
---
license: other
language:
- en
pipeline_tag: text-generation
tags:
- code
---
# Defog SQLCoder
Defog's SQLCoder is a state-of-the-art LLM for converting natural language questions to SQL queries.
[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)
## TL;DR
SQLCoder 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`
SQLCoder is fine-tuned on a base StarCoder model.
## Results on novel datasets not seen in training
| model | perc_correct |
|-|-|
| gpt4-2023-10-04 | 82.0 |
| defog-sqlcoder2 | 77.5 |
| gpt4-2023-08-28 | 74.0 |
| defog-sqlcoder-7b | 71.0 |
| gpt-3.5-2023-10-04 | 66.0 |
| 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, 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.
## 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](https://defog.ai/blog/open-sourcing-sqlcoder2-7b/) and [evaluation framework](https://defog.ai/blog/open-sourcing-sqleval/).
## Results by question category
We 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.
| query_category | gpt-4 | sqlcoder2-15b | sqlcoder-7b | gpt-3.5 | claude-2 | claude-instant | gpt-3 |
|:-----------------|--------:|----------------:|--------------:|----------:|-----------:|-----------------:|--------:|
| date | 72 | 80 | 64 | 68 | 52 | 48 | 32 |
| group_by | 91.4 | 82.9 | 82.9 | 77.1 | 71.4 | 71.4 | 71.4 |
| order_by | 82.9 | 77.1 | 74.3 | 68.6 | 74.3 | 74.3 | 68.6 |
| ratio | 80 | 74.3 | 54.3 | 37.1 | 57.1 | 45.7 | 25.7 |
| join | 82.9 | 74.3 | 74.3 | 71.4 | 65.7 | 62.9 | 57.1 |
| where | 80 | 77.1 | 74.3 | 74.3 | 62.9 | 60 | 54.3 |
## 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](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).
```bash
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](https://defog.ai/sqlcoder-demo), or run SQLCoder in Colab [here](https://colab.research.google.com/drive/13BIKsqHnPOBcQ-ba2p77L5saiepTIwu0#scrollTo=ZpbVgVHMkJvC)
## Hardware Requirements
SQLCoder 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.
## Todo
- [x] Open-source the v1 model weights
- [x] 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 |