Spaces:
Running
Running
More cleaning
Browse files- app.shell.py +0 -143
app.shell.py
DELETED
@@ -1,143 +0,0 @@
|
|
1 |
-
from tiktoken import get_encoding
|
2 |
-
from weaviate_interface import WeaviateClient
|
3 |
-
from prompt_templates import question_answering_prompt_series, question_answering_system
|
4 |
-
from openai_interface import GPT_Turbo
|
5 |
-
from app_features import (convert_seconds, generate_prompt_series, search_result,
|
6 |
-
validate_token_threshold, load_content_cache, load_data)
|
7 |
-
from reranker import ReRanker
|
8 |
-
from loguru import logger
|
9 |
-
import streamlit as st
|
10 |
-
import sys
|
11 |
-
import json
|
12 |
-
import os
|
13 |
-
|
14 |
-
# load environment variables
|
15 |
-
from dotenv import load_dotenv
|
16 |
-
load_dotenv('.env', override=True)
|
17 |
-
|
18 |
-
## PAGE CONFIGURATION
|
19 |
-
st.set_page_config(page_title="Impact Theory",
|
20 |
-
page_icon=None,
|
21 |
-
layout="wide",
|
22 |
-
initial_sidebar_state="auto",
|
23 |
-
menu_items=None)
|
24 |
-
##############
|
25 |
-
# START CODE #
|
26 |
-
##############
|
27 |
-
data_path = 'data/impact_theory_data.json'
|
28 |
-
cache_path = 'data/impact_theory_cache.parquet'
|
29 |
-
data = load_data(data_path)
|
30 |
-
cache = load_content_cache(cache_path)
|
31 |
-
## RETRIEVER
|
32 |
-
client.display_properties.append('summary')
|
33 |
-
## RERANKER
|
34 |
-
|
35 |
-
## LLM
|
36 |
-
|
37 |
-
## ENCODING
|
38 |
-
|
39 |
-
## INDEX NAME
|
40 |
-
|
41 |
-
##############
|
42 |
-
# END CODE #
|
43 |
-
##############
|
44 |
-
data = load_data(data_path)
|
45 |
-
#creates list of guests for sidebar
|
46 |
-
guest_list = sorted(list(set([d['guest'] for d in data])))
|
47 |
-
|
48 |
-
def main():
|
49 |
-
|
50 |
-
with st.sidebar:
|
51 |
-
guest = st.selectbox('Select Guest', options=guest_list, index=None, placeholder='Select Guest')
|
52 |
-
|
53 |
-
st.image('./assets/impact-theory-logo.png', width=400)
|
54 |
-
st.subheader(f"Chat with the Impact Theory podcast: ")
|
55 |
-
st.write('\n')
|
56 |
-
col1, _ = st.columns([7,3])
|
57 |
-
with col1:
|
58 |
-
query = st.text_input('Enter your question: ')
|
59 |
-
st.write('\n\n\n\n\n')
|
60 |
-
|
61 |
-
if query:
|
62 |
-
##############
|
63 |
-
# START CODE #
|
64 |
-
##############
|
65 |
-
|
66 |
-
st.write('Hmmm...this app does not seem to be working yet. Please check back later.')
|
67 |
-
if guest:
|
68 |
-
st.write(f'However, it looks like you selected {guest} as a filter.')
|
69 |
-
# make hybrid call to weaviate
|
70 |
-
hybrid_response = None
|
71 |
-
# rerank results
|
72 |
-
ranked_response = None
|
73 |
-
# validate token count is below threshold
|
74 |
-
# valid_response = validate_token_threshold(ranked_response,
|
75 |
-
# question_answering_prompt_series,
|
76 |
-
# query=query,
|
77 |
-
# tokenizer= # variable from ENCODING,
|
78 |
-
# token_threshold=4000,
|
79 |
-
# verbose=True)
|
80 |
-
##############
|
81 |
-
# END CODE #
|
82 |
-
##############
|
83 |
-
|
84 |
-
# # generate LLM prompt
|
85 |
-
# prompt = generate_prompt_series(query=query, results=valid_response)
|
86 |
-
|
87 |
-
# # prep for streaming response
|
88 |
-
# st.subheader("Response from Impact Theory (context)")
|
89 |
-
# with st.spinner('Generating Response...'):
|
90 |
-
# st.markdown("----")
|
91 |
-
# #creates container for LLM response
|
92 |
-
# chat_container, response_box = [], st.empty()
|
93 |
-
#
|
94 |
-
# # execute chat call to LLM
|
95 |
-
# ##############
|
96 |
-
# # START CODE #
|
97 |
-
# ##############
|
98 |
-
#
|
99 |
-
|
100 |
-
# ##############
|
101 |
-
# # END CODE #
|
102 |
-
# ##############
|
103 |
-
# try:
|
104 |
-
#inserts chat stream from LLM
|
105 |
-
# with response_box:
|
106 |
-
# content = resp.choices[0].delta.content
|
107 |
-
# if content:
|
108 |
-
# chat_container.append(content)
|
109 |
-
# result = "".join(chat_container).strip()
|
110 |
-
# st.write(f'{result}')
|
111 |
-
# except Exception as e:
|
112 |
-
# print(e)
|
113 |
-
# continue
|
114 |
-
# ##############
|
115 |
-
# # START CODE #
|
116 |
-
# ##############
|
117 |
-
# st.subheader("Search Results")
|
118 |
-
# for i, hit in enumerate(valid_response):
|
119 |
-
# col1, col2 = st.columns([7, 3], gap='large')
|
120 |
-
# image = # get thumbnail_url
|
121 |
-
# episode_url = # get episode_url
|
122 |
-
# title = # get title
|
123 |
-
# show_length = # get length
|
124 |
-
# time_string = # convert show_length to readable time string
|
125 |
-
# ##############
|
126 |
-
# # END CODE #
|
127 |
-
# ##############
|
128 |
-
# with col1:
|
129 |
-
# st.write( search_result( i=i,
|
130 |
-
# url=episode_url,
|
131 |
-
# guest=hit['guest'],
|
132 |
-
# title=title,
|
133 |
-
# content=hit['content'],
|
134 |
-
# length=time_string),
|
135 |
-
# unsafe_allow_html=True)
|
136 |
-
# st.write('\n\n')
|
137 |
-
# with col2:
|
138 |
-
# # st.write(f"<a href={episode_url} <img src={image} width='200'></a>",
|
139 |
-
# # unsafe_allow_html=True)
|
140 |
-
# st.image(image, caption=title.split('|')[0], width=200, use_column_width=False)
|
141 |
-
|
142 |
-
if __name__ == '__main__':
|
143 |
-
main()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|