Spaces:
Sleeping
Sleeping
add rag pipeline
Browse files
appStore/__pycache__/target.cpython-310.pyc
ADDED
Binary file (2.89 kB). View file
|
|
appStore/__pycache__/vulnerability_analysis.cpython-310.pyc
ADDED
Binary file (4.74 kB). View file
|
|
appStore/rag.py
ADDED
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import numpy as np
|
3 |
+
import pandas as pd
|
4 |
+
import openai
|
5 |
+
from haystack.schema import Document
|
6 |
+
import streamlit as st
|
7 |
+
from tenacity import retry, stop_after_attempt, wait_random_exponential
|
8 |
+
from huggingface_hub import InferenceClient
|
9 |
+
|
10 |
+
|
11 |
+
# Get openai API key
|
12 |
+
hf_token = os.environ["HF_API_KEY"]
|
13 |
+
|
14 |
+
# define a special function for putting the prompt together (as we can't use haystack)
|
15 |
+
def get_prompt(context, label):
|
16 |
+
base_prompt="Summarize the following context efficiently in bullet points, the less the better - but keep concrete goals. \
|
17 |
+
Summarize only elements of the context that address vulnerability of "+label+" to climate change. \
|
18 |
+
If there is no mention of "+label+" in the context, return: 'No clear references to vulnerability of "+label+" found'. \
|
19 |
+
Do not include an introduction sentence, just the bullet points as per below. \
|
20 |
+
Formatting example: \
|
21 |
+
- Bullet point 1 \
|
22 |
+
- Bullet point 2 \
|
23 |
+
"
|
24 |
+
|
25 |
+
prompt = base_prompt+"; Context: "+context+"; Answer:"
|
26 |
+
|
27 |
+
return prompt
|
28 |
+
|
29 |
+
|
30 |
+
# # exception handling for issuing multiple API calls to openai (exponential backoff)
|
31 |
+
# @retry(wait=wait_random_exponential(min=1, max=60), stop=stop_after_attempt(6))
|
32 |
+
# def completion_with_backoff(**kwargs):
|
33 |
+
# return openai.ChatCompletion.create(**kwargs)
|
34 |
+
|
35 |
+
class ChatCompletionResult:
|
36 |
+
def __init__(self):
|
37 |
+
self.content = ""
|
38 |
+
|
39 |
+
def add_content(self, text):
|
40 |
+
self.content += text
|
41 |
+
|
42 |
+
def get_full_content(self):
|
43 |
+
return self.content.strip()
|
44 |
+
|
45 |
+
def run_query(context, label, model_sel_name):
|
46 |
+
'''
|
47 |
+
Summarize provided test
|
48 |
+
'''
|
49 |
+
chatbot_role = """You are an analyst specializing in climate change impact assessments and producing insights from policy documents."""
|
50 |
+
messages = [{"role": "system", "content": chatbot_role},{"role": "user", "content": get_prompt(context, label)}]
|
51 |
+
|
52 |
+
# Initialize the client, pointing it to one of the available models
|
53 |
+
client = InferenceClient(model_sel_name, token=hf_token)
|
54 |
+
|
55 |
+
# Instantiate ChatCompletion as a generator object (stream is set to True)
|
56 |
+
chat_completion = client.chat.completions.create(
|
57 |
+
messages=messages,
|
58 |
+
stream=True
|
59 |
+
)
|
60 |
+
|
61 |
+
# Create an object to store the full chat completion
|
62 |
+
completion_result = ChatCompletionResult()
|
63 |
+
res_box = st.empty()
|
64 |
+
|
65 |
+
# Iterate through the streamed output
|
66 |
+
for chunk in chat_completion:
|
67 |
+
# Extract the object containing the text
|
68 |
+
if chunk.choices is not None:
|
69 |
+
chunk_message = chunk.choices[0].delta
|
70 |
+
if 'content' in chunk_message:
|
71 |
+
completion_result.add_content(chunk_message['content']) # Store the message
|
72 |
+
# Add the latest text and merge it with all previous
|
73 |
+
result = completion_result.get_full_content()
|
74 |
+
res_box.success(result) # Output to response text box
|
75 |
+
|
76 |
+
# Return the stored chat completion object for later use
|
77 |
+
return completion_result
|
utils/__pycache__/target_classifier.cpython-310.pyc
ADDED
Binary file (3.56 kB). View file
|
|