Spaces:
Running
Running
# import gradio as gr | |
# import requests | |
# import os | |
# ##Bloom | |
# API_URL = "https://api-inference.huggingface.co/models/bigscience/bloom" | |
# HF_TOKEN = "Bloom_Token" | |
# headers = {"Authorization": f"Bearer {HF_TOKEN}"} | |
# def sql_generate(prompt, input_prompt_sql ): | |
# print(f"*****Inside SQL_generate - Prompt is :{prompt}") | |
# print(f"length of input_prompt_sql is {len(input_prompt_sql)}") | |
# print(f"length of prompt is {len(prompt)}") | |
# if len(prompt) == 0: | |
# prompt = input_prompt_sql | |
# json_ = {"inputs": prompt, | |
# "parameters": | |
# { | |
# "top_p": 0.9, | |
# "temperature": 1.1, | |
# "max_new_tokens": 64, | |
# "return_full_text": False, | |
# }, | |
# "options": | |
# {"use_cache": True, | |
# "wait_for_model": True, | |
# },} | |
# response = requests.post(API_URL, headers=headers, json=json_) | |
# print(f"Response is : {response}") | |
# output = response.json() | |
# print(f"output is : {output}") | |
# output_tmp = output[0]['generated_text'] | |
# print(f"output_tmp is: {output_tmp}") | |
# solution = output_tmp.split("\nQ:")[0] | |
# print(f"Final response after splits is: {solution}") | |
# if '\nOutput:' in solution: | |
# final_solution = solution.split("\nOutput:")[0] | |
# print(f"Response after removing output is: {final_solution}") | |
# elif '\n\n' in solution: | |
# final_solution = solution.split("\n\n")[0] | |
# print(f"Response after removing new line entries is: {final_solution}") | |
# else: | |
# final_solution = solution | |
# return final_solution | |
# demo = gr.Blocks() | |
# with demo: | |
# gr.Markdown("<h1><center>Zero Shot SQL by Bloom</center></h1>") | |
# gr.Markdown( | |
# """[BigScienceW Bloom](https://twitter.com/BigscienceW) \n\n Large language models have demonstrated a capability of Zero-Shot SQL generation. Some might say β You can get good results out of LLMs if you know how to speak to them. This space is an attempt at inspecting this behavior/capability in the new HuggingFace BigScienceW [Bloom](https://huggingface.co/bigscience/bloom) model.\n\nThe Prompt length is limited at the API end right now, thus there is a certain limitation in testing Bloom's capability thoroughly.This Space might sometime fail due to inference queue being full and logs would end up showing error as *'queue full, try again later'*, in such cases please try again after few minutes. Please note that, longer prompts might not work as well and the Space could error out with Response code [500] or *'A very long prompt, temporarily not accepting these'* message in the logs. Still iterating over the app, might be able to improve it further soon.. \n\nThis Space is created by [Yuvraj Sharma](https://twitter.com/yvrjsharma) for Gradio EuroPython 2022 Demo.""" | |
# ) | |
# with gr.Row(): | |
# example_prompt = gr.Radio( [ | |
# "Instruction: Given an input question, respond with syntactically correct PostgreSQL\nInput: How many users signed up in the past month?\nPostgreSQL query: ", | |
# "Instruction: Given an input question, respond with syntactically correct PostgreSQL\nInput: Create a query that displays empfname, emplname, deptid, deptname, location from employee table. Results should be in the ascending order based on the empfname and location.\nPostgreSQL query: ", | |
# "Instruction: Given an input question, respond with syntactically correct PostgreSQL. Only use table called 'employees'.\nInput: What is the total salary paid to all the employees?\nPostgreSQL query: ", | |
# "Instruction: Given an input question, respond with syntactically correct PostgreSQL. Only use table called 'employees'.\nInput: List names of all the employees whose name end with 'r'.\nPostgreSQL query: ", | |
# "Instruction: Given an input question, respond with syntactically correct PostgreSQL. Only use table called 'employees'.\nInput: What are the number of employees in each department?\nPostgreSQL query: ", | |
# "Instruction: Given an input question, respond with syntactically correct PostgreSQL. Only use table called 'employees'.\nInput: Select names of all theemployees who have third character in their name as 't'.\nPostgreSQL query: ", | |
# "Instruction: Given an input question, respond with syntactically correct PostgreSQL. Only use table called 'employees'.\nInput: Select names of all the employees who are working under 'Peter'.\nPostgreSQL query: ", ], label= "Choose a sample Prompt") | |
# #with gr.Column: | |
# input_prompt_sql = gr.Textbox(label="Or Write text following the example pattern given below, to get SQL commands...", value="Instruction: Given an input question, respond with syntactically correct PostgreSQL. Use table called 'department'.\nInput: Select names of all the departments in their descending alphabetical order.\nPostgreSQL query: ", lines=6) | |
# with gr.Row(): | |
# generated_txt = gr.Textbox(lines=3) | |
# b1 = gr.Button("Generate SQL") | |
# b1.click(sql_generate,inputs=[example_prompt, input_prompt_sql], outputs=generated_txt) | |
# with gr.Row(): | |
# gr.Markdown("![visitor badge](https://visitor-badge.glitch.me/badge?page_id=europython2022_zero-shot-sql-by-bloom)") | |
# demo.launch(enable_queue=True, debug=True) | |
import gradio as gr | |
gr.Interface.load("models/bigscience/bloom").launch() |