Extremly slow Langchain WebBaseLoader(url).load() in Streamlit

LLMs and AI
,
1

Hello.
I’m testing the behaviour of Langchain WebBaseLoader in Streamlit. In the following code (no Streamlit), the webpage from the specified URL is read in 0.15 seconds.

from langchain_community.document_loaders import WebBaseLoader
import time

tic = time.perf_counter()
loader = WebBaseLoader("https://joint-research-centre.ec.europa.eu/welcome-jec-website/reference-regulatory-framework/renewable-energy-recast-2030-red-ii_en")
tac = time.perf_counter()
data = loader.load()
toc = time.perf_counter()

print(data)
print(f"Loader: {tac - tic:0.3f}, load: {toc - tac:0.3f}")

But, when I just initialize the Streamlit app, the loader.load() takes 10 minutes!

import streamlit as st
from langchain_community.document_loaders import WebBaseLoader
import time

tic = time.perf_counter()
loader = WebBaseLoader("https://joint-research-centre.ec.europa.eu/welcome-jec-website/reference-regulatory-framework/renewable-energy-recast-2030-red-ii_en")
tac = time.perf_counter()
data = loader.load()
toc = time.perf_counter()

st.set_page_config(page_title="Test loader.load() & Streamlit", page_icon="spider")
st.title("Here's where st.title goes.")

print(data)
print(f"Loader: {tac - tic:0.1f}, load: {toc - tac:0.1f}")

The app is being hosted in Firefox, iMac, osx 10.15. Does anyone have an idea what might be going on here?
Cheers