Easy-to-use, offline-first ML experiment management solution.
Project description
tidyexp
Easy-to-use, offline-first ML experiment management solution.
What does it solve?
Other competitors are complex in nature – they have a slight steep learning curve and aren’t often beginner-friendly. Some of them require you to create an account on their platform to get started.
tidyexp aims to provide a free, easy-to-use platform for tracking ML experiment metadata.
Installation
Through pip
:
pip install tidyexp
Usage
Import tidyexp:
import tidyexp
Create a Logger instance with the experiment metadata:
log = tidyexp.Logger(experiment_id="1", experiment_dir=".", time_track=["num_epochs"], stats_track=["mse"], overwrite=True, model_type="torch")
Track experiment metadata in the training loop:
for i in range(epochs):
....
time_dict = {"num_epochs": i}
stats_dict = {"mse": curr_loss}
log.update(time_dict, stats_dict)
Save logs:
log.save()
Load logs:
from tidyexp.load.load_log import load_log, load_stats, load_time
logs = load_log("abcd/logs/log_1.hdf5")
stats = load_stats("abcd/logs/log_1.hdf5", "1")
time_stats = load_time("abcd/logs/log_1.hdf5", "1")
Save model:
log.save_model(model)
Load model:
from tidyexp.load.load_model import load_model
ckpt = load_model("abcd/models/final/final_1.pt", "torch")
Create archive (.zip
):
log.archive("archive")
Upload to Google Drive:
log.upload_gdrive("./credentials.json", "MyExperiment", "archive.zip")
Push to local Git repository:
log.commit("C:\\Users\\ExampleUser\\Experiments", ".\abcd")
License
tidyexp is licensed under the MIT License.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file tidyexp-0.1.0.tar.gz
.
File metadata
- Download URL: tidyexp-0.1.0.tar.gz
- Upload date:
- Size: 9.4 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/4.0.0 CPython/3.9.12
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 1a91f49403da5c511e4c6a933e6031e7b9a6ebbbb4defcef645bff70c6785db5 |
|
MD5 | 4b96320072b3df4c19892b10a63a3666 |
|
BLAKE2b-256 | b71e71783093fe13db2a98cebfe04ffb1fb62c49d7771534a590f361a55cd487 |
File details
Details for the file tidyexp-0.1.0-py3-none-any.whl
.
File metadata
- Download URL: tidyexp-0.1.0-py3-none-any.whl
- Upload date:
- Size: 11.2 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/4.0.0 CPython/3.9.12
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 07d211544bf4d99e450eec0074f6a7a675ee8aed2bc20582c1b3015fc0033f40 |
|
MD5 | 0500e66a5aa8e76161913c41fc950ccc |
|
BLAKE2b-256 | 3366c2ea4f5039404bd8d6107a25dc711c4a02cad5d8d0b56a2dbe88918b8b2c |