Update app.py
Browse files
app.py
CHANGED
@@ -99,7 +99,7 @@ def wrap_model(model, tokenizer):
|
|
99 |
|
100 |
|
101 |
|
102 |
-
def fetch_context(db, model, query,
|
103 |
"""
|
104 |
Perform similarity search and retrieve related context to query.
|
105 |
I have stored large documents in db so I can apply compressor on the set of retrived documents to
|
@@ -140,7 +140,7 @@ def format_context(docs):
|
|
140 |
|
141 |
|
142 |
|
143 |
-
def llm_chain_with_context(model, model_name, query, context, template
|
144 |
"""
|
145 |
Run simple chain with formatted prompt including query and retrieved context and the underlying model to generate a response.
|
146 |
"""
|
@@ -158,11 +158,11 @@ def llm_chain_with_context(model, model_name, query, context, template, logger):
|
|
158 |
return output
|
159 |
|
160 |
|
161 |
-
def generate_response(query, model, template
|
162 |
start_time = time.time()
|
163 |
progress_text = "Loading model. Please wait."
|
164 |
my_bar = st.progress(0, text=progress_text)
|
165 |
-
context = fetch_context(db, model, model_name, query, template
|
166 |
# fill those as appropriate
|
167 |
my_bar.progress(0.1, "Loading Database. Please wait.")
|
168 |
|
@@ -171,7 +171,7 @@ def generate_response(query, model, template, logger):
|
|
171 |
my_bar.progress(0.5, "Running RAG. Please wait.")
|
172 |
|
173 |
my_bar.progress(0.7, "Generating Answer. Please wait.")
|
174 |
-
response = llm_chain_with_context(model, model_name, query, context, template
|
175 |
|
176 |
logger.info(f"Total Execution Time: {time.time() - start_time}")
|
177 |
|
@@ -286,7 +286,7 @@ if __name__=="__main__":
|
|
286 |
if user_question is not None and user_question!="":
|
287 |
with st.chat_message("Human", avatar="π§π»"):
|
288 |
st.write(user_question)
|
289 |
-
response = generate_response(user_question, model, all_templates
|
290 |
with st.chat_message("AI", avatar="ποΈ"):
|
291 |
st.write(response)
|
292 |
|
|
|
99 |
|
100 |
|
101 |
|
102 |
+
def fetch_context(db, model, query, template, use_compressor=True):
|
103 |
"""
|
104 |
Perform similarity search and retrieve related context to query.
|
105 |
I have stored large documents in db so I can apply compressor on the set of retrived documents to
|
|
|
140 |
|
141 |
|
142 |
|
143 |
+
def llm_chain_with_context(model, model_name, query, context, template):
|
144 |
"""
|
145 |
Run simple chain with formatted prompt including query and retrieved context and the underlying model to generate a response.
|
146 |
"""
|
|
|
158 |
return output
|
159 |
|
160 |
|
161 |
+
def generate_response(query, model, template):
|
162 |
start_time = time.time()
|
163 |
progress_text = "Loading model. Please wait."
|
164 |
my_bar = st.progress(0, text=progress_text)
|
165 |
+
context = fetch_context(db, model, model_name, query, template)
|
166 |
# fill those as appropriate
|
167 |
my_bar.progress(0.1, "Loading Database. Please wait.")
|
168 |
|
|
|
171 |
my_bar.progress(0.5, "Running RAG. Please wait.")
|
172 |
|
173 |
my_bar.progress(0.7, "Generating Answer. Please wait.")
|
174 |
+
response = llm_chain_with_context(model, model_name, query, context, template)
|
175 |
|
176 |
logger.info(f"Total Execution Time: {time.time() - start_time}")
|
177 |
|
|
|
286 |
if user_question is not None and user_question!="":
|
287 |
with st.chat_message("Human", avatar="π§π»"):
|
288 |
st.write(user_question)
|
289 |
+
response = generate_response(user_question, model, all_templates)
|
290 |
with st.chat_message("AI", avatar="ποΈ"):
|
291 |
st.write(response)
|
292 |
|