Installation Guide

Use the smallest install that matches your workflow, then validate the console scripts you plan to use.

Choose an install profile

Install name vs import names

Install the distribution as stormlog, then import the Python APIs from stormlog or stormlog.tensorflow:

Task	Use
Install from PyPI	`pip install stormlog`
Launch the TUI	`stormlog`
Import PyTorch APIs	`from stormlog import GPUMemoryProfiler, MemoryTracker`
Import TensorFlow APIs	`from stormlog.tensorflow import TFMemoryProfiler`
Import JAX APIs	`from stormlog.jax import JAXMemoryProfiler`
Run CLI automation	`gpumemprof`, `tfmemprof`, or `jaxmemprof`

Core package

pip install stormlog

Includes:

stormlog
stormlog.tensorflow
stormlog.jax
core telemetry and analysis utilities
CPU-compatible monitoring and tracking

Visualization extras

pip install "stormlog[viz]"

Adds the dependencies used by:

MemoryVisualizer
TUI HTML export from the Visualizations tab
richer plot generation in example scripts

TUI extras

pip install "stormlog[tui,torch]"

Installs the Textual stack plus the current PyTorch runtime dependency required by TUI startup. The stormlog console script is declared by the package; these extras make the app runnable.

Framework extras

pip install "stormlog[torch]"
pip install "stormlog[tf]"
pip install "stormlog[jax]"
pip install "stormlog[all]"

stormlog[all] installs every runtime extra: viz, tui, torch, tf, and jax.

Source checkout

git clone https://github.com/Silas-Asamoah/stormlog.git
cd stormlog
pip install -e .

For a contributor setup with all common extras:

pip install -e ".[dev,test,all]"
pre-commit install

Source-only examples and guides

The examples/ package and the Markdown test guides under the Example Test Guides are only available from a repository checkout. A plain pip install stormlog does not include them.

Verification

Core verification

python3 -c "import stormlog; print(stormlog.__version__)"
gpumemprof --help
gpumemprof info

TensorFlow verification

If you installed the TensorFlow extra:

tfmemprof --help
tfmemprof info

JAX verification

If you installed the JAX extra:

jaxmemprof --help
jaxmemprof info

TUI verification

If you installed the TUI dependencies:

stormlog

If TUI dependencies are missing after a source checkout, reinstall with:

pip install -e ".[tui,torch]"

Platform notes

CPU-only systems

You do not need CUDA to use Stormlog. The CLI, TUI monitoring flows, and CPU profiler helpers remain usable on CPU-only machines. See the CPU Compatibility Guide.

macOS and MPS

The TUI and CLI support Apple Silicon workflows. Use the CLI or TUI monitoring and diagnostics flows even when GPUMemoryProfiler is unavailable on MPS.

Linux and CUDA

Install the correct framework build for your torch.cuda runtime before using GPUMemoryProfiler. For NVIDIA systems that means matching the CUDA toolchain. See the GPU Setup Guide for the full checklist.

Common install failures

`gpumemprof: command not found`

pip install -e .
hash -r
gpumemprof --help

`stormlog: command not found`

pip install -e ".[tui,torch]"
hash -r
stormlog

Missing framework imports

Install the matching extra instead of trying to work around import errors manually:

pip install "stormlog[torch]"
pip install "stormlog[tf]"
pip install "stormlog[jax]"

Next steps

Read the Usage Guide for the Python API and CLI-first workflow.
Read the CLI Guide if you need telemetry, plots, or diagnose bundles.
Read the TUI Guide if you want an interactive terminal workflow.

← Back to main docs