Spaces:
Sleeping
Sleeping
first commit
Browse files- app.py +224 -0
- requirements.txt +0 -0
app.py
ADDED
@@ -0,0 +1,224 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from jamaibase import JamAI, protocol as p
|
3 |
+
import time
|
4 |
+
import os
|
5 |
+
|
6 |
+
# Initialize session state variables
|
7 |
+
def initialize_session_state():
|
8 |
+
if "unique_time" not in st.session_state:
|
9 |
+
st.session_state.unique_time = time.time()
|
10 |
+
if "knowledge_base_exist" not in st.session_state:
|
11 |
+
st.session_state.knowledge_base_exist = False
|
12 |
+
if "chat_history" not in st.session_state:
|
13 |
+
st.session_state.chat_history = []
|
14 |
+
if "api_key" not in st.session_state:
|
15 |
+
st.session_state.api_key = None
|
16 |
+
if "project_id" not in st.session_state:
|
17 |
+
st.session_state.project_id = None
|
18 |
+
if "model" not in st.session_state:
|
19 |
+
st.session_state.model = "ellm/Qwen/Qwen2-7B-Instruct"
|
20 |
+
if "k" not in st.session_state:
|
21 |
+
st.session_state.k = 2
|
22 |
+
if "temperature" not in st.session_state:
|
23 |
+
st.session_state.temperature = 0.01
|
24 |
+
if "top_p" not in st.session_state:
|
25 |
+
st.session_state.top_p = 0.01
|
26 |
+
if "max_tokens" not in st.session_state:
|
27 |
+
st.session_state.max_tokens = 496
|
28 |
+
|
29 |
+
# Function to check if any input is None
|
30 |
+
def check_all_fields_filled(file_upload, api_key, proj_id):
|
31 |
+
return file_upload is not None and api_key and proj_id
|
32 |
+
|
33 |
+
def clear_credentials():
|
34 |
+
st.session_state.api_key = ""
|
35 |
+
st.session_state.project_id = ""
|
36 |
+
st.session_state.unique_time = time.time() # reset the unique time
|
37 |
+
|
38 |
+
# Function to create knowledge base
|
39 |
+
def create_knowledge_base(jamai, file_upload):
|
40 |
+
try:
|
41 |
+
with st.spinner("Creating Knowledge Base..."):
|
42 |
+
knowledge_simple = f"knowledge-simple-{st.session_state.unique_time}"
|
43 |
+
knowledge_table = jamai.create_knowledge_table(
|
44 |
+
p.KnowledgeTableSchemaCreate(
|
45 |
+
id=knowledge_simple,
|
46 |
+
cols=[],
|
47 |
+
embedding_model="ellm/BAAI/bge-m3",
|
48 |
+
)
|
49 |
+
)
|
50 |
+
st.success("Successfully created Knowledge Base")
|
51 |
+
|
52 |
+
# Save PDF to local directory
|
53 |
+
current_dir = os.path.dirname(os.path.abspath(__file__))
|
54 |
+
file_path = os.path.join(current_dir, file_upload.name)
|
55 |
+
with open(file_path, "wb") as f:
|
56 |
+
f.write(file_upload.read())
|
57 |
+
|
58 |
+
# Upload file to knowledge base
|
59 |
+
with st.spinner("Uploading PDF to Knowledge Base..."):
|
60 |
+
response = jamai.upload_file(
|
61 |
+
p.FileUploadRequest(
|
62 |
+
file_path=file_path,
|
63 |
+
table_id=knowledge_simple,
|
64 |
+
)
|
65 |
+
)
|
66 |
+
assert response.ok
|
67 |
+
st.success("Successfully uploaded PDF to Knowledge Base!")
|
68 |
+
st.session_state.knowledge_base_exist = True
|
69 |
+
|
70 |
+
# remove after upload successfully
|
71 |
+
os.remove(file_path)
|
72 |
+
|
73 |
+
return knowledge_simple
|
74 |
+
except Exception as e:
|
75 |
+
clear_credentials()
|
76 |
+
# st.error(f"An error occurred: {str(e)}")
|
77 |
+
st.warning("An error occurred. Please check your credentials and try again.")
|
78 |
+
return None
|
79 |
+
|
80 |
+
# Function to create chat table
|
81 |
+
def create_chat_table(jamai, knowledge_simple):
|
82 |
+
try:
|
83 |
+
with st.spinner("Creating Chat Table..."):
|
84 |
+
table = jamai.create_chat_table(
|
85 |
+
p.ChatTableSchemaCreate(
|
86 |
+
id=f"chat-rag-{st.session_state.unique_time}",
|
87 |
+
cols=[
|
88 |
+
p.ColumnSchemaCreate(id="User", dtype=p.DtypeCreateEnum.str_),
|
89 |
+
p.ColumnSchemaCreate(
|
90 |
+
id="AI",
|
91 |
+
dtype=p.DtypeCreateEnum.str_,
|
92 |
+
gen_config=p.ChatRequest(
|
93 |
+
model=st.session_state.model,
|
94 |
+
messages=[p.ChatEntry.system("You are a concise assistant.")],
|
95 |
+
rag_params=p.RAGParams(
|
96 |
+
table_id=knowledge_simple,
|
97 |
+
k=st.session_state.k,
|
98 |
+
),
|
99 |
+
temperature=st.session_state.temperature,
|
100 |
+
top_p=st.session_state.top_p,
|
101 |
+
max_tokens=st.session_state.max_tokens,
|
102 |
+
).model_dump(),
|
103 |
+
),
|
104 |
+
],
|
105 |
+
)
|
106 |
+
)
|
107 |
+
st.success("Successfully created Chat Table")
|
108 |
+
except Exception as e:
|
109 |
+
clear_credentials()
|
110 |
+
# st.error(f"An error occurred while creating the chat table: {str(e)}")
|
111 |
+
st.warning("An error occurred. Please check your credentials and try again.")
|
112 |
+
|
113 |
+
# Function to ask a question with improved streaming output
|
114 |
+
def ask_question(question):
|
115 |
+
jamai = JamAI(api_key=st.session_state.api_key, project_id=st.session_state.project_id)
|
116 |
+
completion = jamai.add_table_rows(
|
117 |
+
"chat",
|
118 |
+
p.RowAddRequest(
|
119 |
+
table_id=f"chat-rag-{st.session_state.unique_time}",
|
120 |
+
data=[dict(User=question)],
|
121 |
+
stream=True,
|
122 |
+
),
|
123 |
+
)
|
124 |
+
|
125 |
+
full_response = ""
|
126 |
+
|
127 |
+
for chunk in completion:
|
128 |
+
if chunk.output_column_name != "AI":
|
129 |
+
continue
|
130 |
+
if isinstance(chunk, p.GenTableStreamReferences):
|
131 |
+
pass
|
132 |
+
else:
|
133 |
+
full_response += chunk.text
|
134 |
+
yield full_response
|
135 |
+
|
136 |
+
# Main app function
|
137 |
+
def main():
|
138 |
+
st.title("Chat With PDF")
|
139 |
+
|
140 |
+
initialize_session_state()
|
141 |
+
|
142 |
+
# Sidebar for configuration
|
143 |
+
with st.sidebar:
|
144 |
+
st.logo("resource/Jamai-Long-Black-Main.icuEAbYB.svg")
|
145 |
+
login_button = """
|
146 |
+
<a href="https://cloud.jamaibase.com/" style="
|
147 |
+
display: inline-block;
|
148 |
+
padding: 10px 30px;
|
149 |
+
font-size: 1.2rem;
|
150 |
+
font-weight: 600;
|
151 |
+
text-align: center;
|
152 |
+
text-decoration: none;
|
153 |
+
transition: 0.25s;
|
154 |
+
color: #ffffff;
|
155 |
+
background-color: #007bff;
|
156 |
+
border: 1px solid transparent;
|
157 |
+
border-radius: 0.25rem;
|
158 |
+
">
|
159 |
+
Login to your Jamai Base
|
160 |
+
</a>
|
161 |
+
"""
|
162 |
+
|
163 |
+
st.markdown(login_button, unsafe_allow_html=True)
|
164 |
+
|
165 |
+
st.header("Configuration")
|
166 |
+
file_upload = st.file_uploader("Upload your PDF", type=["pdf"], disabled=st.session_state.knowledge_base_exist)
|
167 |
+
api_key = st.text_input('JAMAI API KEY', value=st.session_state.get('api_key', ''), type='password', disabled=st.session_state.knowledge_base_exist)
|
168 |
+
project_id = st.text_input('Project ID', value=st.session_state.get('project_id', ''), disabled=st.session_state.knowledge_base_exist)
|
169 |
+
|
170 |
+
with st.expander("Advanced Settings"):
|
171 |
+
model_options = [
|
172 |
+
'ellm/Qwen/Qwen2-7B-Instruct', 'ellm/Qwen/Qwen2-72B-Instruct',
|
173 |
+
'ellm/meta-llama/Llama-3-8B-Instruct', 'ellm/meta-llama/Llama-3-70B-Instruct',
|
174 |
+
'ellm/meta-llama/Llama-3.1-8B-Instruct', 'ellm/meta-llama/Llama-3.1-70B-Instruct',
|
175 |
+
'ellm/microsoft/Phi-3-mini-128k-Instruct-Int4'
|
176 |
+
'ellm/microsoft/Phi-3-medium-128k-Instruct-Int4',
|
177 |
+
]
|
178 |
+
st.session_state.model = st.selectbox("Model", options=model_options, index=0, disabled=st.session_state.knowledge_base_exist)
|
179 |
+
st.session_state.k = st.slider("k", value=2, min_value=1, max_value=30, step=1, disabled=st.session_state.knowledge_base_exist)
|
180 |
+
st.session_state.max_tokens = st.slider("max tokens", value=496, min_value=96, max_value=960, step=8, disabled=st.session_state.knowledge_base_exist)
|
181 |
+
temperature_options = [str(i/10) for i in range(1,11)]
|
182 |
+
st.session_state.temperature = st.selectbox("temperature", options=temperature_options, format_func=lambda x: float(x), disabled=st.session_state.knowledge_base_exist)
|
183 |
+
top_p_options = [str(i/10) for i in range(1,11)]
|
184 |
+
st.session_state.top_p = st.selectbox("top p", options=top_p_options, format_func=lambda x: float(x), disabled=st.session_state.knowledge_base_exist)
|
185 |
+
|
186 |
+
if st.button("Create Knowledge Base", disabled=st.session_state.knowledge_base_exist):
|
187 |
+
if not check_all_fields_filled(file_upload, api_key, project_id):
|
188 |
+
st.warning("Please provide all required information: PDF file, JAMAI API KEY, and Project ID.")
|
189 |
+
else:
|
190 |
+
st.session_state.api_key = api_key
|
191 |
+
st.session_state.project_id = project_id
|
192 |
+
jamai = JamAI(api_key=api_key, project_id=project_id)
|
193 |
+
knowledge_simple = create_knowledge_base(jamai, file_upload)
|
194 |
+
if knowledge_simple:
|
195 |
+
create_chat_table(jamai, knowledge_simple)
|
196 |
+
|
197 |
+
# Main chat interface
|
198 |
+
st.header("Jamai Base")
|
199 |
+
|
200 |
+
# Display chat history
|
201 |
+
for message in st.session_state.chat_history:
|
202 |
+
with st.chat_message(message["role"]):
|
203 |
+
st.markdown(message["content"])
|
204 |
+
|
205 |
+
# Chat input
|
206 |
+
if question := st.chat_input("Ask a question about your PDF"):
|
207 |
+
if st.session_state.knowledge_base_exist:
|
208 |
+
st.session_state.chat_history.append({"role": "user", "content": question})
|
209 |
+
with st.chat_message("user"):
|
210 |
+
st.markdown(question)
|
211 |
+
|
212 |
+
with st.chat_message("assistant"):
|
213 |
+
message_placeholder = st.empty()
|
214 |
+
full_response = ""
|
215 |
+
for response in ask_question(question):
|
216 |
+
message_placeholder.markdown(response + "▌")
|
217 |
+
full_response = response
|
218 |
+
message_placeholder.markdown(full_response)
|
219 |
+
st.session_state.chat_history.append({"role": "assistant", "content": full_response})
|
220 |
+
else:
|
221 |
+
st.warning("Please upload a PDF and create a Knowledge Base first.")
|
222 |
+
|
223 |
+
if __name__ == "__main__":
|
224 |
+
main()
|
requirements.txt
ADDED
Binary file (2.12 kB). View file
|
|