Streaming
Browse files- app.py +23 -20
- data_preparation.ipynb +1 -1
app.py
CHANGED
@@ -56,7 +56,8 @@ Settings.llm = Groq(
|
|
56 |
model="llama3-8b-8192",
|
57 |
api_key=os.getenv("GROQ_API_KEY"),
|
58 |
max_tokens=6000,
|
59 |
-
context_window=6000
|
|
|
60 |
)
|
61 |
|
62 |
@st.cache_resource
|
@@ -92,13 +93,11 @@ def load_indices():
|
|
92 |
index, vector_retriever, bm25_retriever, hybrid_retriever = load_indices()
|
93 |
|
94 |
# Function to process chat with RAG
|
95 |
-
def chat_with_rag(message, history, retriever):
|
96 |
-
|
97 |
if st.session_state.get('use_rag', True):
|
98 |
nodes = retriever.retrieve(message)
|
99 |
-
# sort nodes by score
|
100 |
nodes = sorted(nodes, key=lambda x: x.score, reverse=True)
|
101 |
-
# nodes up to slider value
|
102 |
nodes = nodes[:st.session_state.get('num_chunks', 1)]
|
103 |
context = "\n\n".join([node.text for node in nodes])
|
104 |
system_prompt = f"""{st.session_state.system_prompt}
|
@@ -108,26 +107,29 @@ def chat_with_rag(message, history, retriever):
|
|
108 |
{context}
|
109 |
"""
|
110 |
|
111 |
-
# Store sources in session state for this message
|
112 |
-
# Calculate the correct message index (total number of messages)
|
113 |
message_index = len(st.session_state.messages)
|
114 |
st.session_state.sources[message_index] = nodes
|
115 |
else:
|
116 |
system_prompt = st.session_state.system_prompt
|
117 |
nodes = []
|
118 |
|
119 |
-
# Prepare messages for the API call
|
120 |
messages = [ChatMessage(role=MessageRole.SYSTEM, content=system_prompt)]
|
121 |
for h in history:
|
122 |
role = MessageRole.ASSISTANT if h["role"] == "assistant" else MessageRole.USER
|
123 |
messages.append(ChatMessage(role=role, content=h["content"]))
|
124 |
messages.append(ChatMessage(role=MessageRole.USER, content=message))
|
125 |
|
126 |
-
#
|
127 |
-
|
128 |
-
|
|
|
|
|
|
|
|
|
129 |
|
130 |
-
return
|
|
|
|
|
131 |
|
132 |
# Move the title to the top, before tabs
|
133 |
st.title("Freud Explorer")
|
@@ -272,14 +274,15 @@ with tab2:
|
|
272 |
|
273 |
with chat_container:
|
274 |
with st.chat_message("assistant"):
|
275 |
-
|
276 |
-
|
277 |
-
|
278 |
-
|
279 |
-
|
280 |
-
|
281 |
-
|
282 |
-
|
|
|
283 |
|
284 |
st.rerun()
|
285 |
|
|
|
56 |
model="llama3-8b-8192",
|
57 |
api_key=os.getenv("GROQ_API_KEY"),
|
58 |
max_tokens=6000,
|
59 |
+
context_window=6000,
|
60 |
+
stream=True # Enable streaming
|
61 |
)
|
62 |
|
63 |
@st.cache_resource
|
|
|
93 |
index, vector_retriever, bm25_retriever, hybrid_retriever = load_indices()
|
94 |
|
95 |
# Function to process chat with RAG
|
96 |
+
def chat_with_rag(message, history, retriever, response_placeholder):
|
97 |
+
"""Modified to handle streaming"""
|
98 |
if st.session_state.get('use_rag', True):
|
99 |
nodes = retriever.retrieve(message)
|
|
|
100 |
nodes = sorted(nodes, key=lambda x: x.score, reverse=True)
|
|
|
101 |
nodes = nodes[:st.session_state.get('num_chunks', 1)]
|
102 |
context = "\n\n".join([node.text for node in nodes])
|
103 |
system_prompt = f"""{st.session_state.system_prompt}
|
|
|
107 |
{context}
|
108 |
"""
|
109 |
|
|
|
|
|
110 |
message_index = len(st.session_state.messages)
|
111 |
st.session_state.sources[message_index] = nodes
|
112 |
else:
|
113 |
system_prompt = st.session_state.system_prompt
|
114 |
nodes = []
|
115 |
|
|
|
116 |
messages = [ChatMessage(role=MessageRole.SYSTEM, content=system_prompt)]
|
117 |
for h in history:
|
118 |
role = MessageRole.ASSISTANT if h["role"] == "assistant" else MessageRole.USER
|
119 |
messages.append(ChatMessage(role=role, content=h["content"]))
|
120 |
messages.append(ChatMessage(role=MessageRole.USER, content=message))
|
121 |
|
122 |
+
# Stream the response
|
123 |
+
response_text = ""
|
124 |
+
for response in Settings.llm.stream_chat(messages):
|
125 |
+
if response.delta is not None:
|
126 |
+
response_text += response.delta
|
127 |
+
# Update the placeholder with the accumulated text
|
128 |
+
response_placeholder.markdown(response_text + "▌")
|
129 |
|
130 |
+
# Remove the cursor and return the complete response
|
131 |
+
response_placeholder.markdown(response_text)
|
132 |
+
return response_text
|
133 |
|
134 |
# Move the title to the top, before tabs
|
135 |
st.title("Freud Explorer")
|
|
|
274 |
|
275 |
with chat_container:
|
276 |
with st.chat_message("assistant"):
|
277 |
+
# Create a placeholder for the streaming response
|
278 |
+
response_placeholder = st.empty()
|
279 |
+
response = chat_with_rag(
|
280 |
+
prompt,
|
281 |
+
st.session_state.messages[:-1],
|
282 |
+
hybrid_retriever if st.session_state.use_rag else None,
|
283 |
+
response_placeholder
|
284 |
+
)
|
285 |
+
st.session_state.messages.append({"role": "assistant", "content": response})
|
286 |
|
287 |
st.rerun()
|
288 |
|
data_preparation.ipynb
CHANGED
@@ -66,7 +66,7 @@
|
|
66 |
"path=\"txt\\Freud_Complete_en.txt\"\n",
|
67 |
"\n",
|
68 |
"if os.path.exists(path):\n",
|
69 |
-
" print(load_txt(path)[:1000])"
|
70 |
]
|
71 |
},
|
72 |
{
|
|
|
66 |
"path=\"txt\\Freud_Complete_en.txt\"\n",
|
67 |
"\n",
|
68 |
"if os.path.exists(path):\n",
|
69 |
+
" print(load_txt(path)[:1000]) "
|
70 |
]
|
71 |
},
|
72 |
{
|