Update README.md
Browse files
README.md
CHANGED
|
@@ -15,4 +15,43 @@ tags:
|
|
| 15 |
metrics:
|
| 16 |
- accuracy
|
| 17 |
- character
|
| 18 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
metrics:
|
| 16 |
- accuracy
|
| 17 |
- character
|
| 18 |
+
library_name: transformers
|
| 19 |
+
---
|
| 20 |
+
|
| 21 |
+
Industry standard text to sql generation with high accuracy.
|
| 22 |
+
|
| 23 |
+
sample code to start with:
|
| 24 |
+
|
| 25 |
+
import torch
|
| 26 |
+
from transformers import T5Tokenizer, T5ForConditionalGeneration
|
| 27 |
+
|
| 28 |
+
# Initialize the tokenizer from Hugging Face Transformers library
|
| 29 |
+
tokenizer = T5Tokenizer.from_pretrained('anilajax/text2sql_industry_standard')
|
| 30 |
+
|
| 31 |
+
# Load the model
|
| 32 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 33 |
+
model = T5ForConditionalGeneration.from_pretrained('anilajax/text2sql_industry_standard')
|
| 34 |
+
model = model.to(device)
|
| 35 |
+
model.eval()
|
| 36 |
+
|
| 37 |
+
def generate_sql(input_prompt):
|
| 38 |
+
# Tokenize the input prompt
|
| 39 |
+
inputs = tokenizer(input_prompt, padding=True, truncation=True, return_tensors="pt").to(device)
|
| 40 |
+
|
| 41 |
+
# Forward pass
|
| 42 |
+
with torch.no_grad():
|
| 43 |
+
outputs = model.generate(**inputs, max_length=512)
|
| 44 |
+
|
| 45 |
+
# Decode the output IDs to a string (SQL query in this case)
|
| 46 |
+
generated_sql = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
| 47 |
+
|
| 48 |
+
return generated_sql
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
input_prompt = "provide count of students where class = 10"
|
| 52 |
+
|
| 53 |
+
generated_sql = generate_sql(input_prompt)
|
| 54 |
+
|
| 55 |
+
print(f"The generated SQL query is: {generated_sql}")
|
| 56 |
+
## expected output - SELECT COUNT(*) FROM students WHERE class = 10
|
| 57 |
+
|