TinyExp 0.0.4

A minimalist Python project for deep learning experiment management. It uses Ray for core distributed environment and backend setup, and provides basic, no-frills tracking for models, optimizers, and LR schedulers.

TinyExp

A simple Python project for deep learning experiment management.

TinyExp lets you launch experiments with one click: the file you edit becomes the entrypoint to your experiment.

Usage

pip install tinyexp

1. Run mnist example(By default, It will trained on CPU)

import tinyexp
from tinyexp.examples.mnist_exp import Exp, store_and_run_exp
store_and_run_exp(Exp)

2. or:

python tinyexp/examples/mnist_exp.py

3. Run mnist example with overridden config:

python tinyexp/examples/mnist_exp.py dataloader_cfg.train_batch_size_per_device=16

see all available configs:

python tinyexp/examples/mnist_exp.py mode=help

see all available configs with overridden configs:

python tinyexp/examples/mnist_exp.py mode=help dataloader_cfg.train_batch_size_per_device=16

More Examples

1. ImageNet ResNet-50 Example with Extremely Fast Data Loading (By default, all available GPUs will be used.)

# export IMAGENET_HOME=yours_imagenet_dir

import tinyexp
from tinyexp.examples.resnet_exp import ResNetExp, store_and_run_exp
store_and_run_exp(ResNetExp)

Develop

1. prepare env

# 1. clone repo
git clone https://github.com/HKUST-SAIL/tinyexp.git
# 2. Set Up Your Development Environment, This will also generate your `uv.lock` file
make install
source .venv/bin/activate

2. After development, checking whether the code is standardized

# Initially, the CI/CD pipeline might be failing due to formatting issues. To resolve those run:
uv run pre-commit run -a

3. Release a new version

# One command to auto-generate CHANGELOG entry (from commits since last tag),
# bump version, test, tag, publish and push:
make release VERSION=0.0.4
# (The script uses git-cliff; if git-cliff is not installed, it falls back to `uvx --from git-cliff git-cliff`.)

License

MIT License. See LICENSE.