Skip to content

JupyterLab Essentials

The next-generation Jupyter interface: notebooks, text editors, terminals, and data viewers in one extensible workspace. Still uses the same .ipynb format — everything from Jupyter applies.

Table of Contents

Installation / Setup

pip install jupyterlab        # or: uv add jupyterlab
jupyter lab                   # launch at localhost:8888/lab

Same kernels as classic notebooks (see jupyter.md for ipykernel install).

pip install jupyterlab-git jupyterlab-code-formatter jupyterlab-variableinspector
jupyter labextension list     # see installed extensions

Core Syntax

The workspace

Area What it is
Left sidebar File browser, running kernels/terminals, extension manager, git
Main area Tabbed notebooks, editors, terminals, viewers
Right sidebar Inspector (when enabled)
Top menu File, Edit, View, Run, Kernel, Tabs, Settings, Help

Everything is a dockable pane — drag tabs to split side-by-side or stacked.

Keyboard shortcuts (Command mode = Esc)

Keys Action
Ctrl+Shift+P / Cmd+Shift+P command palette (discover everything)
Ctrl+B / Cmd+B toggle left sidebar
Esc then A / B insert cell above / below
Shift+Enter run cell, advance
Ctrl+Enter run cell, stay
Esc then DD delete cell
Ctrl+S save notebook
Ctrl+Shift+D duplicate notebook/editor tab
Ctrl+Shift+F find across open files

Common Operations

File browser & multiple views

# Drag a notebook to the right half to watch results while editing a script on the left
# Right-click a file -> "Open With" -> Notebook / Editor / JSON viewer

Right-click → Open in new tab lets you compare a notebook with its exported script.

Terminals & consoles

# Left sidebar -> other -> Terminal (full shell next to your notebooks)
# Consoles give a REPL attached to the same kernel as a notebook

# In a notebook/console: check the kernel env
import sys; print(sys.executable)

Data & image viewers

# Double-click a CSV in the file browser -> built-in data viewer
# Double-click an image / JSON / Parquet preview
# Variable inspector extension shows DataFrame shapes as you run

Git integration

  • Install jupyterlab-git → Git pane in the sidebar: stage, commit, push, view diffs of .ipynb and files.
  • For clean diffs use nbstripout (strips outputs on commit).

Search & navigation

# Ctrl+Shift+F  -> search across all open files
# Click an error traceback -> jumps to the file/line
# Outline pane (right side in notebook): headings + function list

Practical Examples

1. Notebook + script workflow

# 1. Write tested logic in a .py module
# 2. In JupyterLab open the module, right-click -> "Create Console for Editor"
#    (or use %autoreload) to test interactively
# 3. Keep notebooks as thin, runnable narratives

2. Side-by-side experiment comparison

# Open two notebooks that train model A and model B
# Drag one to the right split -> run both, compare loss curves live

3. Custom extensions

pip install jupyterlab-plotly     # interactive plotly in notebook output
pip install ipympl                # interactive matplotlib widgets
pip install jupyterlab-tabular-data-editor   # spreadsheet-like editing

Common Patterns

# Reproducible lab workspace: document the environment
# requirements:  pip freeze > requirements.txt  (see pip.md / uv.md)

# Server on a remote box, browse locally
#   jupyter lab --no-browser --port=8888
#   then: ssh -L 8888:localhost:8888 user@server  (tunnel)

Best Practices

  • Use the command palette (Ctrl+Shift+P) as your main way to learn the UI.
  • Organize the left sidebar: file browser, running kernels, git.
  • Trust but verify kernels — a notebook runs whatever kernel you chose in the top-right.
  • Restart & Run All before sharing any notebook.
  • Save early, save often; JupyterLab autosaves by default (check Settings).
  • Enable autosave + git pane so notebook edits are never lost.
  • Use extensions sparingly — each adds startup weight and update friction.

Common Mistakes

Mistake Consequence Fix
Running the wrong kernel (again) Wrong env imports Check top-right kernel name
Too many open kernels RAM exhaustion Shut down unused kernels (Running pane)
Editing .py without reload Stale behavior in notebook %autoreload 2 or restart kernel
Dragging cells across notebooks Shared state confusion Copy/paste deliberately
Ignoring the file browser for data Can't find CSVs Pin the data folder; use search
Installing extensions in wrong env Extensions don't load Install where JupyterLab is installed

Troubleshooting

Symptom Cause Fix
jupyter lab opens old notebook UI Wrong command/alias Use jupyter lab not jupyter notebook
Extension missing from UI Not installed/restart needed pip install; rebuild/restart lab
Port 8888 busy Another server jupyter lab --port 8890
Kernel dies on huge DataFrame Memory Free memory, close other kernels, chunk reads
Git pane absent Extension not installed pip install jupyterlab-git; refresh

Official Documentation

Last Updated

2026-09-06 — Initial version.