HridaAIofficial
commited on
Commit
•
3f101c7
1
Parent(s):
3199c14
Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,131 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
language:
|
4 |
+
- en
|
5 |
+
library_name: transformers
|
6 |
+
pipeline_tag: text2text-generation
|
7 |
+
tags:
|
8 |
+
- sql
|
9 |
+
- t2sql
|
10 |
+
- text2sql
|
11 |
+
---
|
12 |
+
|
13 |
+
The **Hrida-T2SQL-3B-V0.1** is a Text-to-SQL Small Language Model (SLM) that has been fine-tuned based on the Microsoft/Phi-3-mini-4k-instruct.
|
14 |
+
|
15 |
+
For full details of this model please read our [blog post](https://www.hridaai.com/blog/t2sql).
|
16 |
+
|
17 |
+
## Prompt Template
|
18 |
+
|
19 |
+
```txt
|
20 |
+
### Instruction:
|
21 |
+
Provide the system prompt.
|
22 |
+
|
23 |
+
### Dialect:
|
24 |
+
Specify the SQL dialect (e.g., MySQL, PostgreSQL, SQL Server, etc.).
|
25 |
+
|
26 |
+
### Context:
|
27 |
+
Provide the database schema including table names, column names, and data types.
|
28 |
+
|
29 |
+
### Input:
|
30 |
+
User's query.
|
31 |
+
|
32 |
+
### Response:
|
33 |
+
Expected SQL query output based on the input and context.
|
34 |
+
|
35 |
+
```
|
36 |
+
|
37 |
+
- **Instruction (System Prompt)**: This guides the model on processing input to generate the SQL query response effectively.
|
38 |
+
- **Dialect (Optional)**: Specify the SQL variant the model should use to ensure the generated query conforms to the correct syntax.
|
39 |
+
- **Context**: Provide the database schema to the model for generating accurate SQL queries.
|
40 |
+
- **Input**: Provide the user query for the model to comprehend and transform into an SQL query.
|
41 |
+
- **Response**: Expected output from the model.
|
42 |
+
|
43 |
+
|
44 |
+
## Chat Prompt Template
|
45 |
+
|
46 |
+
```txt
|
47 |
+
<s>
|
48 |
+
<|system|>
|
49 |
+
{ Instruction / System Prompt }
|
50 |
+
<|user|>
|
51 |
+
{ Context / User Query } <|end|>
|
52 |
+
<|assistant|>
|
53 |
+
```
|
54 |
+
|
55 |
+
## Run the Model
|
56 |
+
|
57 |
+
### Using Transformers
|
58 |
+
|
59 |
+
```python
|
60 |
+
import torch
|
61 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
62 |
+
|
63 |
+
# Define the model and tokenizer
|
64 |
+
model_id = "HridaAI/Hrida-T2SQL-3B-V0.1"
|
65 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
|
66 |
+
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16, trust_remote_code=True)
|
67 |
+
|
68 |
+
# Define the context and prompt
|
69 |
+
prompt = """
|
70 |
+
Answer to the query will be in the form of an SQL query.
|
71 |
+
### Context: CREATE TABLE Employees (
|
72 |
+
EmployeeID INT PRIMARY KEY,
|
73 |
+
FirstName VARCHAR(50),
|
74 |
+
LastName VARCHAR(50),
|
75 |
+
Age INT,
|
76 |
+
DepartmentID INT,
|
77 |
+
Salary DECIMAL(10, 2),
|
78 |
+
DateHired DATE,
|
79 |
+
Active BOOLEAN,
|
80 |
+
FOREIGN KEY (DepartmentID) REFERENCES Departments(DepartmentID)
|
81 |
+
);
|
82 |
+
|
83 |
+
CREATE TABLE Departments (
|
84 |
+
DepartmentID INT PRIMARY KEY,
|
85 |
+
DepartmentName VARCHAR(100),
|
86 |
+
Location VARCHAR(100)
|
87 |
+
);
|
88 |
+
### Input: Write a SQL query to select all the employees who are active.
|
89 |
+
### Response:
|
90 |
+
"""
|
91 |
+
# Prepare the input
|
92 |
+
messages = [{"role": "user", "content": prompt}]
|
93 |
+
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt", add_generation_prompt=True)
|
94 |
+
|
95 |
+
# Generate the output
|
96 |
+
outputs = model.generate(inputs, max_length=300)
|
97 |
+
print(tokenizer.decode(outputs[0]))
|
98 |
+
|
99 |
+
|
100 |
+
```
|
101 |
+
|
102 |
+
### Using MLX
|
103 |
+
|
104 |
+
```python
|
105 |
+
from mlx_lm import generate, load
|
106 |
+
|
107 |
+
model,tokenizer = load("HridaAI/Hrida-T2SQL-3B-V0.1")
|
108 |
+
|
109 |
+
prompt = """
|
110 |
+
Answer to the quey will be in the form of SQL query.
|
111 |
+
### Context: CREATE TABLE Employees (
|
112 |
+
EmployeeID INT PRIMARY KEY,
|
113 |
+
FirstName VARCHAR(50),
|
114 |
+
LastName VARCHAR(50),
|
115 |
+
Age INT,
|
116 |
+
DepartmentID INT,
|
117 |
+
Salary DECIMAL(10, 2),
|
118 |
+
DateHired DATE,
|
119 |
+
Active BOOLEAN,
|
120 |
+
FOREIGN KEY (DepartmentID) REFERENCES Departments(DepartmentID)
|
121 |
+
);
|
122 |
+
|
123 |
+
CREATE TABLE Departments (
|
124 |
+
DepartmentID INT PRIMARY KEY,
|
125 |
+
DepartmentName VARCHAR(100),
|
126 |
+
Location VARCHAR(100)
|
127 |
+
); ### Input: Write a SQL query to select all the employees who are active. ### Response:"""
|
128 |
+
|
129 |
+
response = generate(model=model,tokenizer=tokenizer,prompt=prompt, verbose=True)
|
130 |
+
|
131 |
+
```
|