Update app.py

app.py

@@ -15,7 +15,7 @@ import pdfkit
 from paddleocr import PaddleOCR
 import fitz  
 import asyncio
-from langchain_nomic.embeddings import NomicEmbeddings
+# from langchain_nomic.embeddings import NomicEmbeddings
 
 # initialise LLM model
 llm_groq = ChatGroq(
@@ -26,13 +26,7 @@ llm_groq = ChatGroq(
 anonymizer = PresidioReversibleAnonymizer(analyzed_fields=['PERSON', 'EMAIL_ADDRESS', 'PHONE_NUMBER', 'IBAN_CODE', 'CREDIT_CARD', 'CRYPTO', 'IP_ADDRESS', 'LOCATION', 'DATE_TIME', 'NRP', 'MEDICAL_LICENSE', 'URL', 'US_BANK_NUMBER', 'US_DRIVER_LICENSE', 'US_ITIN', 'US_PASSPORT', 'US_SSN'], faker_seed=18)
 
 # initalise nomic embedding model
-embeddings = NomicEmbeddings(model="nomic-embed-text-v1.5")
-
-def embed_text(text):
-    if len(text.split()) <= 50:
-        return embeddings.embed_query(text)
-    else:
-        return embeddings.embed_document(text)
+# embeddings = NomicEmbeddings(model="nomic-embed-text-v1.5")
 
 def extract_text_from_pdf(file_path):
     pdf = PyPDF2.PdfReader(file_path)
@@ -154,7 +148,7 @@ async def on_chat_start():
     # without splitting into chunks
     # {
     # Create a Chroma vector store
-    # embeddings = OllamaEmbeddings(model="nomic-embed-text")
+    embeddings = OllamaEmbeddings(model="nomic-embed-text")
     docsearch = await cl.make_async(Chroma.from_texts)(
         [anonymized_text], embeddings, metadatas=[{"source": "0-pl"}]
     )
@@ -198,7 +192,7 @@ async def main(message: cl.Message):
     # Call the chain with user's message content
     res = await chain.ainvoke(message.content, callbacks=[cb])
     answer = anonymizer.deanonymize(
-        "ok"+res["answer"]
+        res["answer"]
     )  
     text_elements = []