UPDATE: Flashrank

functions.py

@@ -11,7 +11,8 @@ from langchain_core.chat_history import BaseChatMessageHistory
 from langchain.storage import InMemoryStore
 from langchain.docstore.document import Document
 from langchain_huggingface import HuggingFaceEmbeddings
-from langchain_cohere import CohereRerank
+from langchain.retrievers import ContextualCompressionRetriever
+from langchain.retrievers.document_compressors import FlashrankRerank
 from supabase.client import create_client
 from qdrant_client import QdrantClient
 from langchain_groq import ChatGroq
@@ -118,13 +119,11 @@ def addDocuments(text: str, vectorstore: str):
         global embeddings
         global store
         parentSplitter = RecursiveCharacterTextSplitter(
-            chunk_size = 2400,
-            chunk_overlap = 240,
+            chunk_size = 2100,
             add_start_index = True
         )
         childSplitter = RecursiveCharacterTextSplitter(
             chunk_size = 300,
-            chunk_overlap = 90,
             add_start_index = True
         )
         texts = [Document(page_content = text)]
@@ -191,9 +190,9 @@ def answerQuery(query: str, vectorstore: str, llmModel: str = "llama3-70b-8192")
         vectorstore=vectorstore,
         docstore=store,
         child_splitter=RecursiveCharacterTextSplitter(),
-
+        search_kwargs={"k": 20}
     )
-    compressor = CohereRerank(model="rerank-english-v3.0")
+    compressor = FlashrankRerank()
     retriever = ContextualCompressionRetriever(
         base_compressor=compressor, base_retriever=retriever
     )

requirements.txt

@@ -1,5 +1,6 @@
 huggingface-hub
 fastapi
+flashrank
 gradio
 langchain
 langchain-community