Skip to content
Sentiment Studio AI Platform & Data · Streamlit 1.66.1 Launch on AWS

Configuration

This page lists the Sentiment Studio account, the files it writes, and the limits on input. The starter does not take a provider key or an external database.

Where things live

Path What it is
sushi-streamlit.service The application service.
sushi-streamlit-firstboot.service First-boot setup. It creates the account file.
127.0.0.1:8501 Sign-in page. Open it through a Session Manager tunnel.
/home/ubuntu/.config/sushi-streamlit/access.json Username, password, and the other identity fields. Mode 0600.
/var/lib/streamlit Application state to include in a backup.
/opt/streamlit/sample.csv Six synthetic CSV examples.
~/streamlit-results JSON files written by sushi-streamlit text and sushi-streamlit csv.

Do not delete the identity files to force a new setup. The published guide says that does not trigger a safe reset.

Sign-in account

Read access.json as ubuntu when you need the username and password. Type them into the form. Leave every other field in the file unchanged. Sign out in the browser to clear that browser session. The file remains the account for the next session.

Keep the file out of command arguments, logs, screenshots, recordings, and support mail.

Text

The documented text example is helpful excellent service. From the shell:

sushi-streamlit text 'helpful excellent service' \
  --output ~/streamlit-results/my-text.json

Pick a new --output path for every run. A new name leaves earlier results in place.

CSV

A CSV must be:

  • UTF-8
  • exactly the header row_id,text
  • 1 to 12 data rows
  • at most 16 KiB
  • one distinct row_id per row, using only letters, digits, underscores, and hyphens
sushi-streamlit csv /opt/streamlit/sample.csv \
  --output ~/streamlit-results/my-csv.json
sushi-streamlit verify ~/streamlit-results/my-csv.json

verify checks the saved rows in the JSON file you pass. It does not replace looking at the labels and totals in the browser.

Listen address

Forward workstation port 8501 to the instance's port 8501, and open http://127.0.0.1:8501 on the workstation. Keep both ends on localhost. The published guide says not to expose the application port to the internet.

In the page, Download result JSON uses a loopback link. That link reaches the result while the server still retains it.

Last updated November 1, 2026 · Suggest an improvement