Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,9 +1,12 @@
|
|
| 1 |
import os
|
|
|
|
|
|
|
| 2 |
|
| 3 |
# Install necessary libraries using os.system
|
| 4 |
os.system("pip install --upgrade pip")
|
| 5 |
os.system("pip install streamlit llama-cpp-agent huggingface_hub trafilatura beautifulsoup4 requests duckduckgo-search googlesearch-python")
|
| 6 |
|
|
|
|
| 7 |
try:
|
| 8 |
from llama_cpp import Llama
|
| 9 |
from llama_cpp_agent.providers import LlamaCppPythonProvider
|
|
@@ -19,10 +22,21 @@ try:
|
|
| 19 |
from utils import CitingSources
|
| 20 |
from settings import get_context_by_model, get_messages_formatter_type
|
| 21 |
except ImportError as e:
|
| 22 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
|
| 24 |
import logging
|
| 25 |
-
import streamlit as st
|
| 26 |
from huggingface_hub import hf_hub_download
|
| 27 |
|
| 28 |
# Download the models
|
|
@@ -43,7 +57,9 @@ hf_hub_download(
|
|
| 43 |
)
|
| 44 |
|
| 45 |
# Function to respond to user messages
|
| 46 |
-
def respond(message, history,
|
|
|
|
|
|
|
| 47 |
chat_template = get_messages_formatter_type(model)
|
| 48 |
llm = Llama(
|
| 49 |
model_path=f"models/{model}",
|
|
@@ -80,7 +96,6 @@ def respond(message, history, model, system_message, max_tokens, temperature, to
|
|
| 80 |
settings.temperature = temperature
|
| 81 |
settings.top_k = top_k
|
| 82 |
settings.top_p = top_p
|
| 83 |
-
|
| 84 |
settings.max_tokens = max_tokens
|
| 85 |
settings.repeat_penalty = repeat_penalty
|
| 86 |
|
|
@@ -139,51 +154,19 @@ def respond(message, history, model, system_message, max_tokens, temperature, to
|
|
| 139 |
outputs += "\n".join(citing_sources.sources)
|
| 140 |
yield outputs
|
| 141 |
|
| 142 |
-
|
| 143 |
-
st.title("Llama-CPP-Agent Chatbot with Web Search")
|
| 144 |
-
|
| 145 |
-
# Sidebar for settings
|
| 146 |
-
st.sidebar.title("Settings")
|
| 147 |
-
model = st.sidebar.selectbox(
|
| 148 |
-
"Model",
|
| 149 |
-
[
|
| 150 |
-
'Mistral-7B-Instruct-v0.3-Q6_K.gguf',
|
| 151 |
-
'mixtral-8x7b-instruct-v0.1.Q5_K_M.gguf',
|
| 152 |
-
'Meta-Llama-3-8B-Instruct-Q6_K.gguf'
|
| 153 |
-
]
|
| 154 |
-
)
|
| 155 |
-
system_message = st.sidebar.text_area("System message", value=web_search_system_prompt)
|
| 156 |
-
max_tokens = st.sidebar.slider("Max tokens", min_value=1, max_value=4096, value=2048, step=1)
|
| 157 |
-
temperature = st.sidebar.slider("Temperature", min_value=0.1, max_value=1.0, value=0.45, step=0.1)
|
| 158 |
-
top_p = st.sidebar.slider("Top-p", min_value=0.1, max_value=1.0, value=0.95, step=0.05)
|
| 159 |
-
top_k = st.sidebar.slider("Top-k", min_value=0, max_value=100, value=40, step=1)
|
| 160 |
-
repeat_penalty = st.sidebar.slider("Repetition penalty", min_value=0.0, max_value=2.0, value=1.1, step=0.1)
|
| 161 |
-
|
| 162 |
-
# Chat history
|
| 163 |
-
if "history" not in st.session_state:
|
| 164 |
-
st.session_state.history = []
|
| 165 |
|
| 166 |
-
|
| 167 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 168 |
|
| 169 |
if st.button("Send"):
|
| 170 |
-
|
| 171 |
-
response
|
| 172 |
-
|
| 173 |
-
history,
|
| 174 |
-
|
| 175 |
-
system_message,
|
| 176 |
-
max_tokens,
|
| 177 |
-
temperature,
|
| 178 |
-
top_p,
|
| 179 |
-
top_k,
|
| 180 |
-
repeat_penalty
|
| 181 |
-
)
|
| 182 |
-
|
| 183 |
-
for res in response:
|
| 184 |
-
st.session_state.history.append((message, res))
|
| 185 |
-
st.text_area("Chat", value=f"You: {message}\nBot: {res}", height=300)
|
| 186 |
-
|
| 187 |
-
# Display chat history
|
| 188 |
-
for user_msg, bot_msg in st.session_state.history:
|
| 189 |
-
st.text_area("Chat", value=f"You: {user_msg}\nBot: {bot_msg}", height=300)
|
|
|
|
| 1 |
import os
|
| 2 |
+
import logging
|
| 3 |
+
import streamlit as st
|
| 4 |
|
| 5 |
# Install necessary libraries using os.system
|
| 6 |
os.system("pip install --upgrade pip")
|
| 7 |
os.system("pip install streamlit llama-cpp-agent huggingface_hub trafilatura beautifulsoup4 requests duckduckgo-search googlesearch-python")
|
| 8 |
|
| 9 |
+
# Attempt to import all required modules
|
| 10 |
try:
|
| 11 |
from llama_cpp import Llama
|
| 12 |
from llama_cpp_agent.providers import LlamaCppPythonProvider
|
|
|
|
| 22 |
from utils import CitingSources
|
| 23 |
from settings import get_context_by_model, get_messages_formatter_type
|
| 24 |
except ImportError as e:
|
| 25 |
+
st.error(f"Error importing modules: {e}")
|
| 26 |
+
if 'utils' in str(e):
|
| 27 |
+
st.warning("Mocking utils.CitingSources")
|
| 28 |
+
class CitingSources:
|
| 29 |
+
sources = []
|
| 30 |
+
|
| 31 |
+
if 'settings' in str(e):
|
| 32 |
+
st.warning("Mocking settings functions")
|
| 33 |
+
def get_context_by_model(model):
|
| 34 |
+
return 4096
|
| 35 |
+
|
| 36 |
+
def get_messages_formatter_type(model):
|
| 37 |
+
return MessagesFormatterType.BASIC
|
| 38 |
|
| 39 |
import logging
|
|
|
|
| 40 |
from huggingface_hub import hf_hub_download
|
| 41 |
|
| 42 |
# Download the models
|
|
|
|
| 57 |
)
|
| 58 |
|
| 59 |
# Function to respond to user messages
|
| 60 |
+
def respond(message, history, system_message, temperature, top_p, top_k, repeat_penalty):
|
| 61 |
+
model = "mixtral-8x7b-instruct-v0.1.Q5_K_M.gguf"
|
| 62 |
+
max_tokens = 3000
|
| 63 |
chat_template = get_messages_formatter_type(model)
|
| 64 |
llm = Llama(
|
| 65 |
model_path=f"models/{model}",
|
|
|
|
| 96 |
settings.temperature = temperature
|
| 97 |
settings.top_k = top_k
|
| 98 |
settings.top_p = top_p
|
|
|
|
| 99 |
settings.max_tokens = max_tokens
|
| 100 |
settings.repeat_penalty = repeat_penalty
|
| 101 |
|
|
|
|
| 154 |
outputs += "\n".join(citing_sources.sources)
|
| 155 |
yield outputs
|
| 156 |
|
| 157 |
+
st.title("Novav2 Web Engine")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 158 |
|
| 159 |
+
message = st.text_input("Enter your message:")
|
| 160 |
+
history = st.session_state.get("history", [])
|
| 161 |
+
system_message = st.text_area("System message", value=web_search_system_prompt)
|
| 162 |
+
temperature = st.slider("Temperature", min_value=0.1, max_value=1.0, value=0.45, step=0.1)
|
| 163 |
+
top_p = st.slider("Top-p", min_value=0.1, max_value=1.0, value=0.95, step=0.05)
|
| 164 |
+
top_k = st.slider("Top-k", min_value=0, max_value=100, value=40, step=1)
|
| 165 |
+
repeat_penalty = st.slider("Repetition penalty", min_value=0.0, max_value=2.0, value=1.1, step=0.1)
|
| 166 |
|
| 167 |
if st.button("Send"):
|
| 168 |
+
response_generator = respond(message, history, system_message, temperature, top_p, top_k, repeat_penalty)
|
| 169 |
+
for response in response_generator:
|
| 170 |
+
st.write(response)
|
| 171 |
+
history.append((message, response))
|
| 172 |
+
st.session_state["history"] = history
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|