| import openai | |
| import gradio as gr | |
| import os | |
| #OpenAi call | |
| def gpt3(texts): | |
| openai.api_key = os.environ["Secret"] | |
| response = openai.Completion.create( | |
| engine="text-davinci-003", | |
| prompt= texts, | |
| temperature=0, | |
| max_tokens=750, | |
| top_p=1, | |
| frequency_penalty=0.0, | |
| presence_penalty=0.0, | |
| stop = (";", "/*", "</code>") | |
| ) | |
| x = response.choices[0].text | |
| return x | |
| # Function to elicit sql response from model | |
| def greet(prompt): | |
| txt= (f'''/*Prompt: {prompt}*/ \n —-SQL Code:\n''') | |
| sql = gpt3(txt) | |
| return sql | |
| #Code to set up Gradio UI | |
| iface = gr.Interface(greet, inputs = ["text"], outputs = "text",title="Natural Language to SQL", description="Enter any prompt and get a SQL statement back! For better results, give it more context") | |
| iface.launch() |