Update app.py

app.py

@@ -27,7 +27,7 @@ lang_api_key = os.getenv("lang_api_key")
 os.environ["LANGCHAIN_TRACING_V2"] = "true"
 os.environ["LANGCHAIN_ENDPOINT"] = "https://api.langchain.plus"
 os.environ["LANGCHAIN_API_KEY"] = lang_api_key
-os.environ["LANGCHAIN_PROJECT"] = "Chat with multiple PDFs"
+os.environ["LANGCHAIN_PROJECT"] = "Lithuanian_Law_RAG_QA"
 
 
 
@@ -43,21 +43,6 @@ def create_retriever_from_chroma(vectorstore_path="docs/chroma/", search_type='m
         encode_kwargs=encode_kwargs
     )
 
-
-    llm = llamacpp.LlamaCpp(
-        model_path='qwen2-0_5b-instruct-q4_0.gguf',
-        n_gpu_layers=0,
-        temperature=0.1,
-        top_p=0.9,
-        n_ctx=22000,
-        n_batch=2000,
-        max_tokens=200,
-        repeat_penalty=1.7,
-        last_n_tokens_size = 1500,
-        # callback_manager=callback_manager,
-        verbose=False,
-    )
-
     
     # Check if vectorstore exists
     if os.path.exists(vectorstore_path) and os.listdir(vectorstore_path):
@@ -107,7 +92,7 @@ def main():
     st.markdown("Hi, I am Qwen, chat mmodel, based on respublic of Lithuania law document. Write you question and press enter to start chat.")
 
     
-    retriever = create_retriever_from_chroma(vectorstore_path="docs/chroma/", search_type='mmr', k=12, chunk_size=120, chunk_overlap=20)
+    retriever = create_retriever_from_chroma(vectorstore_path="docs/chroma/", search_type='mmr', k=12, chunk_size=300, chunk_overlap=20)
     if user_question := st.text_input("Ask a question about your documents:"):
         handle_userinput(user_question,retriever)
         
@@ -153,9 +138,9 @@ def create_conversational_rag_chain(retriever):
     callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
 
     llm = llamacpp.LlamaCpp(
-        model_path = "qwen2-0_5b-instruct-q4_0.gguf",
+        model_path = "qwen2-0_5b-instruct-q8_0.gguf",
         n_gpu_layers=0,
-        temperature=0.4,
+        temperature=0.2,
         top_p=0.9,
         n_ctx=22000,
         n_batch=2000,