[Python] LangChain: 'dict' object has no attribute 'replace' when using Chroma retriever

I am working on a chatbot using LangChain, ChromaDB, and Hugging Face models. However, when I try to run my script, I get the following error:

import os
import dotenv
from langchain.prompts import (
ChatPromptTemplate,
SystemMessagePromptTemplate,
HumanMessagePromptTemplate,
PromptTemplate
)
from langchain.schema.runnable import RunnablePassthrough
from langchain_core.output_parsers import StrOutputParser
from langchain_chroma import Chroma
from huggingface_hub import InferenceClient
from langchain_huggingface import HuggingFaceEmbeddings, HuggingFaceEndpoint


dotenv.load_dotenv()
HUGGINGFACEHUB_API_TOKEN = os.getenv("HUGGINGFACEHUB_API_TOKEN")


review_template_str = """You are an AI assistant for Edu Overseas.
You ONLY answer questions related to universities, admission processes, study
programs, visa requirements, and scholarships.
If a question is NOT related to these topics, firmly reply:
'I can only answer questions related to university admissions and study abroad
processes.'

Context:
{context}

Question:
{question}
"""

review_system_prompt = SystemMessagePromptTemplate(
prompt=PromptTemplate(
input_variables=["context"],
template=review_template_str,
)
)
review_human_prompt = HumanMessagePromptTemplate(
prompt=PromptTemplate(
input_variables=["question"],
template="{question}",
)
)

messages = [review_system_prompt, review_human_prompt]
review_prompt_template = ChatPromptTemplate(
input_variables=["context", "question"],
messages=messages,
)

# Initializing Hugging Face embeddings
embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-
base-v2")

# Load Chroma database
reviews_vector_db = Chroma(
persist_directory="chroma_data/", # Path to your Chroma database
embedding_function=embeddings,
)
reviews_retriever = reviews_vector_db.as_retriever(k=10)
relevant documents

# Initializing Hugging Face Inference Client
llm = HuggingFaceEndpoint(
repo_id="tiiuae/falcon-7b-instruct",
token=HUGGINGFACEHUB_API_TOKEN,
max_new_tokens = 200
)

# Define the chatbot pipeline
review_chain = (
{
"context": reviews_retriever | (lambda docs: "\n".join([doc.page_content for
doc in docs])),
"question": RunnablePassthrough()
}
| review_prompt_template
| llm

| StrOutputParser()
)


__all__ = ["review_chain", "reviews_vector_db"]


This is my create_retriever.py code:

import dotenv
import pandas as pd
from langchain_community.vectorstores import Chroma
from langchain_huggingface import HuggingFaceEmbeddings
from langchain.schema import Document


dotenv.load_dotenv()


CSV_PATH = "D:/BotUi/AI_LLM/data/Book1.csv"
CHROMA_PATH = "chroma_data/"

try:
df = pd.read_csv(CSV_PATH)

df.columns = df.columns.str.strip()


required_columns = [
"University", "Course Name", "Website URL", "Campus", "Duration",
"Intakes", "Entry Requirements", "IELTS Score", "IELTS No Band Less Than",
"Tuition Fee", "Remarks"
]
missing_columns = [col for col in required_columns if col not in df.columns]
if missing_columns:
raise ValueError(f" Missing columns in CSV: {', '.join(missing_columns)}")

print("All required columns found. Proceeding with data processing...")

df = df.dropna(subset=["Course Name", "University"])


documents = []
for _, row in df.iterrows():

content = f"""
University: {row['University']}
Course: {row['Course Name']}
Website: {row.get('Website URL', 'Not available')}
Campus: {row.get('Campus', 'Not specified')}
Duration: {row.get('Duration', 'Not specified')}
Intakes: {row.get('Intakes', 'Not specified')}
Entry Requirements: {row.get('Entry Requirements', 'Not specified')}
IELTS Score: {row.get('IELTS Score', 'Not specified')}
Minimum IELTS Band: {row.get('IELTS No Band Less Than', 'Not specified')}
Tuition Fee: {row.get('Tuition Fee', 'Not specified')}
Remarks: {row.get('Remarks', 'None')}
"""

metadata = {col: row[col] for col in df.columns}

documents.append(Document(page_content=content, metadata=metadata))


embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-
base-v2")


vector_db = Chroma.from_documents(
documents=documents, embedding=embeddings, persist_directory=CHROMA_PATH
)

print(f"Successfully stored {len(documents)} course records in ChromaDB.")

except Exception as e:
print(f" Error: {e}")


This is my uni.py:

from chatbot import review_chain

def ask_question():
print("DEBUG: ask_question() function started!") # Debug print
while True:
query = input("Ask a university-related question (or type 'exit' to quit): ")
if query.lower() == "exit":
break
try:
print(f"DEBUG: Sending query -> {query}")
response = review_chain.invoke({"question": query})
print(f"\nDEBUG: {response}\n")
print("\nAI Response:", response, "\n")
except Exception as e:
print(f" Error: {e}\n")

if __name__ == "__main__":
print("DEBUG: Starting uni.py script") # Debug print
ask_question()


What could be causing this 'dict' object has no attribute 'replace' error in LangChain?

Is there a better approach to processing the retrieved documents before passing them to the model?

Any help or guidance would be appreciated!

Continue reading...