Answer questions, commit work in progress

BuildingAChainlitApp.md

@@ -133,6 +133,8 @@ Simply put, this downloads the file as a temp file, we load it in with `TextFile
 
 Why do we want to support streaming? What about streaming is important, or useful?
 
+It helps reduce the user's wait time until the first token appears, making chatbot apps seem more responsive.
+
 ## On Chat Start:
 
 The next scope is where "the magic happens". On Chat Start is when a user begins a chat session. This will happen whenever a user opens a new chat window, or refreshes an existing chat window.
@@ -175,6 +177,8 @@ Now, we'll save that into our user session!
 
 Why are we using User Session here? What about Python makes us need to use this? Why not just store everything in a global variable?
 
+It separates a user's info from other users, which is important in order to keep track of conversational context, uploaded documents, and other session-specific information.
+
 ## On Message
 
 First, we load our chain from the user session:

aimakerspace/qa_pipeline.py

@@ -1,4 +1,5 @@
 from rank_bm25 import BM25Plus
+from langchain.vectorstores import Qdrant
 
 from .openai_utils.prompts import (
     SystemRolePrompt
@@ -17,14 +18,23 @@ def bm25plus_rerank(corpus, query, initial_ranking, top_n=3):
     ranked_indices = [initial_ranking[i] for i in bm25_scores.argsort()[::-1]]    
     return ranked_indices[:top_n]
 
+def search_by_text(qdrant: Qdrant, query_text: str, k: int, return_as_text: bool = False) -> List[Tuple[str, float]]:
+    results = qdrant.similarity_search_with_score(query_text, k)
+    if return_as_text:
+        return [result[0].page_content for result in results]
+    return [(result[0].page_content, result[1]) for result in results]
+
 
 class RetrievalAugmentedQAPipeline:
-    def __init__(self, llm: ChatOpenAI(), vector_db_retriever: VectorDatabase) -> None:
+    def __init__(self, llm: ChatOpenAI(), vector_db_retriever) -> None:
         self.llm = llm
         self.vector_db_retriever = vector_db_retriever
 
     async def arun_pipeline(self, user_query: str):
-        context_list = self.vector_db_retriever.search_by_text(user_query, k=4)
+        if type(self.vector_db_retriever == "Qdrant"):
+            context_list = search_by_text(self.vector_db_retriever,user_query, k=4)
+        else:
+            context_list = self.vector_db_retriever.search_by_text(user_query, k=4)
 
         context_prompt = ""
         for context in context_list:
@@ -46,8 +56,10 @@ class RerankedQAPipeline(RetrievalAugmentedQAPipeline):
     async def arun_pipeline(self, user_query: str, rerank: bool=False) -> str:
         # Retrieve the top 10 results. Either return the top 3, or rerank with BM25 and then return 
         # the new top 3
-        context_list = self.vector_db_retriever.search_by_text(user_query, k=10)
-
+        if type(self.vector_db_retriever == "Qdrant"):
+            context_list = search_by_text(self.vector_db_retriever,user_query, k=10)
+        else:
+            context_list = self.vector_db_retriever.search_by_text(user_query, k=10)
         # Convert from tuples to strings
         context_list_str = [context_list[i][0] for i in range(len(context_list))]