Spaces:
Sleeping
Sleeping
File size: 4,416 Bytes
3b44354 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 |
from openai import OpenAI
import gradio as gr
import requests
from datetime import date
from test_web_rag import get_docs_from_web
import json
import os
from dotenv import load_dotenv
load_dotenv()
# Replace with your key
client = OpenAI()
you_key = os.getenv("YOU_API_KEY")
def get_ai_snippets_for_query(query):
headers = {"X-API-Key": you_key}
params = {"query": query}
return requests.get(
f"https://api.ydc-index.io/search?query={query}",
params=params,
headers=headers,
).json().get('hits')
def get_web_search_you(query):
docs = get_ai_snippets_for_query(query)
markdown = ""
for doc in docs:
for key, value in doc.items():
if key == 'snippets':
markdown += f"{key}:\n"
for snippet in value:
markdown += f"- {snippet}\n"
else:
markdown += f"{key}: {value}\n"
markdown += "\n"
return markdown
def predict(message, history, _n_web_search, _strategy):
# docs = get_web_search_you(message)
with open('history.json', mode='a', encoding='utf-8') as f:
json.dump(history, f)
docs = get_docs_from_web(message, history[-1:], _n_web_search, _strategy)
partial_message = ''
information = ''
for doc in docs:
if isinstance(doc, dict):
information = doc.get('data')
else:
partial_message = partial_message + doc
yield partial_message
system_prompt = """
You are an advanced chatbot.
Today's date - {date}
When answering a question, adhere to the following revised rules:
- The "Information for reference" data is provided in the chunks with each chunk having its own source as url.
- Generate human-like text in response to input, reflecting your status as a sophisticated language model.
- Abstain from offering any health or medical advice and ensure all responses maintain this guideline strictly.
- Format all responses in markdown format consistently throughout interactions.
- Must cite sources from the information at the conclusion of your response using properly titled references, but only if the information you provided comes from sources that can be cited.
Information for reference:
"{context}"
Your answer should be structured in markdown as follows:
<Answer>
**Sources**:
Include this section only if the provided information contains sources. If sources are included, list them as follows:
- [Title of Source 1](URL to Source 1)
- [Title of Source 2](URL to Source 2)
... as needed. If no sources are provided, do not include this section in answer.
""".format(context=information, question=message, date=date.today().strftime('%B %d, %Y'))
history_openai_format = [{"role": "system", "content": system_prompt}]
for human, assistant in history:
history_openai_format.append({"role": "user", "content": human})
history_openai_format.append({"role": "assistant", "content": assistant})
history_openai_format.append({"role": "user", "content": message})
# print(history_openai_format)
response = client.chat.completions.create(model='gpt-4-turbo',
messages=history_openai_format,
temperature=0.5,
max_tokens=1000,
top_p=0.5,
stream=True)
partial_message += '\n\n'
for chunk in response:
if chunk.choices[0].delta.content is not None:
partial_message = partial_message + chunk.choices[0].delta.content
yield partial_message
n_web_search = gr.Slider(1, 10, value=3, step=1, label="Web searches",
info="Choose between 1 and 10 number of web searches to do. Remember more the web searches more it will take time to reply.")
strategy = gr.Radio(["Deep", "Normal"], label="Strategy", value="Normal",
info="Select web search analysis type. Please keep in mind that deep analysis will take more time than normal analysis.")
app = gr.ChatInterface(predict, additional_inputs=[n_web_search, strategy])
app.queue(default_concurrency_limit=5)
app.launch(debug=True, share=False)
|