| Name | probjax JSON |
| Version |
0.1.0
JSON |
| download |
| home_page | None |
| Summary | Jax library for probabilistic computations |
| upload_time | 2024-11-07 13:57:47 |
| maintainer | None |
| docs_url | None |
| author | None |
| requires_python | >=3.9 |
| license | MIT |
| keywords |
probabilistic
jax
computation
|
| VCS |
|
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# Probjax
Probabilistic computation in JAX. This library is under active development and is not yet ready for use. It aims to provide a simple and flexible way to build probabilistic models and perform inference in then. It provides the following set of tools:
- **Core**: A set of core function transformations and primitives useful for building probabilistic models.
- **Traceing**: Tracing and manipulation of function traces. (Very incomplete)
- **Automatic inversion**: Automatic inversion of functions. (Rather complete, with some limitations)
- **Automatic log_prob**: Automatic computation of log-probabilities (Rather incomplete). Automatic computation of log-probabilities of transformed distributions (Rather complete, through automatic inversion and logdet).
- **Distributions**: A set of distributions with support for sampling, log-probability and more.
- **Inference**: Some inference algorithms. (incomplete)
- **Neural networks**: Some neural network layers and models. Based on [Haiku](www.github.com/deepmind/dm-haiku). Here a classical layers as Transformers, Resnets or U-Nets. But also specialised layers for normalising flows, such as coupling layers, autoregressive layers, etc. (complete)
- **Utilities**: Some utilities for numerical computation i.e. odeint, sdeint, etc. (complete)
## Installation
Probjax can be installed using pip:
```bash
pip install -e probjax
```
Additionally, you can install benchmark scripts using:
```bash
pip install -e probjax/scoresbibm
```
Raw data
{
"_id": null,
"home_page": null,
"name": "probjax",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.9",
"maintainer_email": null,
"keywords": "probabilistic, jax, computation",
"author": null,
"author_email": "Manuel Gloeckler <manuel.gloeckler@uni-tuebingen.de>",
"download_url": "https://files.pythonhosted.org/packages/37/56/2c55b9f006fb25098a591158d3ba2ca485efefb0a065107975ece1f60d47/probjax-0.1.0.tar.gz",
"platform": null,
"description": "# Probjax\n\nProbabilistic computation in JAX. This library is under active development and is not yet ready for use. It aims to provide a simple and flexible way to build probabilistic models and perform inference in then. It provides the following set of tools:\n- **Core**: A set of core function transformations and primitives useful for building probabilistic models.\n - **Traceing**: Tracing and manipulation of function traces. (Very incomplete)\n - **Automatic inversion**: Automatic inversion of functions. (Rather complete, with some limitations)\n - **Automatic log_prob**: Automatic computation of log-probabilities (Rather incomplete). Automatic computation of log-probabilities of transformed distributions (Rather complete, through automatic inversion and logdet).\n- **Distributions**: A set of distributions with support for sampling, log-probability and more.\n- **Inference**: Some inference algorithms. (incomplete)\n- **Neural networks**: Some neural network layers and models. Based on [Haiku](www.github.com/deepmind/dm-haiku). Here a classical layers as Transformers, Resnets or U-Nets. But also specialised layers for normalising flows, such as coupling layers, autoregressive layers, etc. (complete)\n- **Utilities**: Some utilities for numerical computation i.e. odeint, sdeint, etc. (complete)\n\n## Installation\n\nProbjax can be installed using pip:\n\n```bash\npip install -e probjax\n```\n\nAdditionally, you can install benchmark scripts using:\n\n```bash\npip install -e probjax/scoresbibm\n```\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "Jax library for probabilistic computations",
"version": "0.1.0",
"project_urls": null,
"split_keywords": [
"probabilistic",
" jax",
" computation"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "82768d2741f656db182b1a0545b5fc2a8338d953bfcaa4225b524211c9dfd478",
"md5": "a041d4445f544595ff7fd3fb0f0b30b7",
"sha256": "ae43839f886aae73a2e44a2d79f9009d4f37e4b49f47fa484325e243ed2ed5e5"
},
"downloads": -1,
"filename": "probjax-0.1.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "a041d4445f544595ff7fd3fb0f0b30b7",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.9",
"size": 152203,
"upload_time": "2024-11-07T13:57:45",
"upload_time_iso_8601": "2024-11-07T13:57:45.965918Z",
"url": "https://files.pythonhosted.org/packages/82/76/8d2741f656db182b1a0545b5fc2a8338d953bfcaa4225b524211c9dfd478/probjax-0.1.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "37562c55b9f006fb25098a591158d3ba2ca485efefb0a065107975ece1f60d47",
"md5": "e6185bbaa74a27023c3797f365f3e8a3",
"sha256": "3ae99d53894e0288f51c162031a7095e46dbafbdc0baef05d27ac22a070b90c9"
},
"downloads": -1,
"filename": "probjax-0.1.0.tar.gz",
"has_sig": false,
"md5_digest": "e6185bbaa74a27023c3797f365f3e8a3",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.9",
"size": 117176,
"upload_time": "2024-11-07T13:57:47",
"upload_time_iso_8601": "2024-11-07T13:57:47.779921Z",
"url": "https://files.pythonhosted.org/packages/37/56/2c55b9f006fb25098a591158d3ba2ca485efefb0a065107975ece1f60d47/probjax-0.1.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-11-07 13:57:47",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "probjax"
}
