Tool for reading and writing datasets of tensors with MessagePack and Lightning Memory-Mapped Database (LMDB)
Project description
ml-pyxis
Tool for reading and writing datasets of tensors (numpy.ndarray) with MessagePack and Lightning Memory-Mapped Database (LMDB).
Example
import numpy as np
import pyxis as px
# Create data
nb_samples = 10
X = np.ones((nb_samples, 2, 2), dtype=np.float32)
y = np.arange(nb_samples, dtype=np.uint8)
# Write
db = px.Writer(dirpath='data', map_size_limit=1)
db.put_samples('input', X, 'target', y)
db.close()
# Read
db = px.Reader(dirpath='data')
sample = db[0]
db.close()
print(sample)
{'input': array([[ 1., 1.], [ 1., 1.]], dtype=float32), 'target': array(0, dtype=uint8)}
More examples can be found in the examples/ directory.
Installation
The installation instructions are generic and should work on most operating systems that support the prerequisites.
ml-pyxis requires Python version 2.7, 3.4, 3.5, or 3.6. We recommend installing ml-pyxis, as well as all prerequisites, in a virtual environment via virtualenv.
Prerequisites
The following Python packages are required to use ml-pyxis:
lmdb - Universal Python binding for the LMDB ‘Lightning’ Database
msgpack - MessagePack implementation for Python (binary serialisation)
NumPy - N-dimensional array object and tools for operating on them
six - A Python 2 and 3 compatibility library
Please refer to the individual packages for more information about additional dependencies and how to install them for your operating system.
Bleeding-edge installation
To install the latest version of ml-pyxis, use the following command:
pip install --upgrade https://github.com/vicolab/ml-pyxis/archive/master.zip
Add the --user tag if you want to install the package in your home directory.
Notice
The previous LMDB-only API has been deprecated in favour of a combination between LMDB and msgpack. The old version can be installed by using the following commit hash with pip:
pip install --upgrade git+git://github.com/vicolab/ml-pyxis.git@787c3484e3121f2767b254fc41be091d0a3e0cf0
Development installation
ml-pyxis can be installed from source in such a way that any changes to your local copy will take effect without having to reinstall the package. Start by making a copy of the repository:
git clone https://github.com/vicolab/ml-pyxis.git
Next, enter the directory and install ml-pyxis in development mode by issuing the following command:
cd ml-pyxis
python setup.py develop
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
File details
Details for the file ml-pyxis-0.4.dev0.tar.gz
.
File metadata
- Download URL: ml-pyxis-0.4.dev0.tar.gz
- Upload date:
- Size: 11.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.4.1 importlib_metadata/4.6.0 pkginfo/1.7.0 requests/2.25.1 requests-toolbelt/0.9.1 tqdm/4.61.1 CPython/3.9.5
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 |
6c5d30cad51ef58ca44916b7fb22e12b10a529f944871e924c3db3b3b3581866
|
|
MD5 |
87886ac6733bb56d6be9bbda1ef70329
|
|
BLAKE2b-256 |
9f6625ee8d6b704f81b4d145fd977dd76b8dbffafbfff60378b0198b760647be
|