Update app.py

app.py

@@ -1,10 +1,10 @@
 import streamlit as st
 from dotenv import load_dotenv
 from PyPDF2 import PdfReader
-from langchain.text_splitter import CharacterTextSplitter
+from langchain.text_splitter import CharacterTextSplitter,RecursiveCharacterTextSplitter
 from langchain.llms import CTransformers  # For loading transformer models.
 from langchain.embeddings import OpenAIEmbeddings, HuggingFaceInstructEmbeddings
-from langchain.vectorstores import FAISS
+from langchain.vectorstores import FAISS, Chroma
 from langchain.embeddings import HuggingFaceEmbeddings  # General embeddings from HuggingFace models.
 from langchain.chat_models import ChatOpenAI
 from langchain.memory import ConversationBufferMemory
@@ -21,12 +21,17 @@ def get_pdf_text(pdf_docs):
 
 
 def get_text_chunks(text):
-    text_splitter = CharacterTextSplitter(
-        separator="\n",
-        chunk_size=1000,
-        chunk_overlap=200,
-        length_function=len
+    text_splitter = RecursiveCharacterTextSplitter(
+        chunk_size = 300,
+        chunk_overlap = 20,
+        length_function= len
     )
+    # text_splitter = CharacterTextSplitter(
+    #     separator="\n",
+    #     chunk_size=1000,
+    #     chunk_overlap=200,
+    #     length_function=len
+    # )
     chunks = text_splitter.split_text(text)
     return chunks
 
@@ -37,14 +42,17 @@ def get_vectorstore(text_chunks):
                                        model_kwargs={'device': 'cpu'})
     # embeddings = OpenAIEmbeddings()
     # embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-xl")
-    vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
+    # vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
+    vectorstore = Chroma.from_texts(texts=text_chunks, embedding=embeddings)
+
     return vectorstore
 
 
 def get_conversation_chain(vectorstore):
     # llm = ChatOpenAI()
     # llm = HuggingFaceHub(repo_id="google/flan-t5-xxl", model_kwargs={"temperature":0.5, "max_length":512})
-    llm = CTransformers(model="llama-2-7b-chat.ggmlv3.q2_K.bin", model_type="llama")
+    config = {'max_new_tokens': 2048}
+    llm = CTransformers(model="llama-2-7b-chat.ggmlv3.q2_K.bin", model_type="llama", config=config)
     memory = ConversationBufferMemory(
         memory_key='chat_history', return_messages=True)
     conversation_chain = ConversationalRetrievalChain.from_llm(