---
base_model: deepseek-ai/deepseek-coder-6.7b-instruct
tags:
  - instruct
  - finetune
library_name: transformers
license: cc-by-sa-4.0
pipeline_tag: text-generation
---

# **Natural-SQL-7B by ChatDB**
## Natural-SQL-7B is a model with very strong performance in Text-to-SQL instructions, has an excellent understanding of complex questions, and outperforms models of the same size in its space.

<img src="https://cdn-uploads.huggingface.co/production/uploads/648a374f00f7a3374ee64b99/hafdsfrFCqrVbATIzV_EN.png" width="600">

[ChatDB.ai](https://chatdb.ai) | [Notebook](https://github.com/cfahlgren1/natural-sql/blob/main/natural-sql-7b.ipynb) | [Twitter](https://twitter.com/calebfahlgren)

# **Benchmarks**
### *Results on Novel Datasets not trained on via SQL-Eval*
<img src="https://cdn-uploads.huggingface.co/production/uploads/648a374f00f7a3374ee64b99/5ynfoKPzI3_-WasQQt7qR.png" width="800">

<em>Big thanks to the [defog](https://huggingface.co/defog) team for open sourcing [sql-eval](https://github.com/defog-ai/sql-eval)</em>👏

Natural-SQL also can handle complex, compound questions that other models typically struggle with. There is a more detailed writeup Here is a write up, small test done [here](https://chatdb.ai/post/naturalsql-vs-sqlcoder-for-text-to-sql).
# Usage

Make sure you have the correct version of the transformers library installed:

```sh
pip install transformers==4.35.2
```

### Loading the Model

Use the following Python code to load the model:

```python
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("chatdb/natural-sql-7b")
model = AutoModelForCausalLM.from_pretrained(
    "chatdb/natural-sql-7b",
    device_map="auto",
    torch_dtype=torch.float16,
)
```

### **License**

The model weights are licensed under `CC BY-SA 4.0`, with extra guidelines for responsible use expanded from the original model's [Deepseek](https://github.com/deepseek-ai/deepseek-coder/blob/main/LICENSE-MODEL) license. 
You're free to use and adapt the model, even commercially. 
If you alter the weights, such as through fine-tuning, you must publicly share your changes under the same `CC BY-SA 4.0` license.


### Generating SQL

```python
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
generated_ids = model.generate(
    **inputs,
    num_return_sequences=1,
    eos_token_id=100001,
    pad_token_id=100001,
    max_new_tokens=400,
    do_sample=False,
    num_beams=1,
)

outputs = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
print(outputs[0].split("```sql")[-1])
```
# Prompt Template

```
# Task 
Generate a SQL query to answer the following question: `{natural language question}`

### PostgreSQL Database Schema 
The query will run on a database with the following schema: 

<SQL Table DDL Statements>

# SQL 
Here is the SQL query that answers the question: `{natural language question}` 
'''sql
```


# Example SQL Output

### Example Schemas

```sql
CREATE TABLE users (
        user_id SERIAL PRIMARY KEY,
        username VARCHAR(50) NOT NULL,
        email VARCHAR(100) NOT NULL,
        password_hash TEXT NOT NULL,
        created_at TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP
    );
CREATE TABLE projects (
    project_id SERIAL PRIMARY KEY,
    project_name VARCHAR(100) NOT NULL,
    description TEXT,
    start_date DATE,
    end_date DATE,
    owner_id INTEGER REFERENCES users(user_id)
);
CREATE TABLE tasks (
    task_id SERIAL PRIMARY KEY,
    task_name VARCHAR(100) NOT NULL,
    description TEXT,
    due_date DATE,
    status VARCHAR(50),
    project_id INTEGER REFERENCES projects(project_id)
);
CREATE TABLE taskassignments (
    assignment_id SERIAL PRIMARY KEY,
    task_id INTEGER REFERENCES tasks(task_id),
    user_id INTEGER REFERENCES users(user_id),
    assigned_date DATE NOT NULL DEFAULT CURRENT_TIMESTAMP
);
CREATE TABLE comments (
    comment_id SERIAL PRIMARY KEY,
    content TEXT NOT NULL,
    created_at TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
    task_id INTEGER REFERENCES tasks(task_id),
    user_id INTEGER REFERENCES users(user_id)
);
```
### Example SQL Outputs

**Question**: **Show me the day with the most users joining**
```sql
SELECT created_at::DATE AS day, COUNT(*) AS user_count
FROM users
GROUP BY day
ORDER BY user_count DESC
LIMIT 1;
```
**Question**: **Show me the project that has a task with the most comments**
```sql
SELECT p.project_name, t.task_name, COUNT(c.comment_id) AS comment_count
FROM projects p
JOIN tasks t ON p.project_id = t.project_id
JOIN comments c ON t.task_id = c.task_id
GROUP BY p.project_name, t.task_name
ORDER BY comment_count DESC
LIMIT 1;
```

**Question**: **What is the ratio of users with gmail addresses vs without?**
```sql
SELECT 
    SUM(CASE WHEN email ILIKE '%@gmail.com%' THEN 1 ELSE 0 END)::FLOAT / NULLIF(SUM(CASE WHEN email NOT ILIKE '%@gmail.com%' THEN 1 ELSE 0 END), 0) AS gmail_ratio
FROM 
    users;
```