rax

Rax is a Learning-to-Rank library written in JAX.

https://github.com/google/rax

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

✓
CITATION.cff file
Found CITATION.cff file
✓
codemeta.json file
Found codemeta.json file
✓
.zenodo.json file
Found .zenodo.json file
○
DOI references
○
Academic publication links
○
Committers with academic emails
○
Institutional organization owner
○
JOSS paper metadata
○
Scientific vocabulary similarity
Low similarity (14.3%) to scientific vocabulary

Keywords from Contributors

jax

Last synced: 7 months ago · JSON representation ·

Repository

Rax is a Learning-to-Rank library written in JAX.

Basic Info

Host: GitHub
Owner: google
License: apache-2.0
Language: Python
Default Branch: main
Homepage: https://rax.readthedocs.io
Size: 280 KB

Statistics

Stars: 331
Watchers: 4
Forks: 12
Open Issues: 2
Releases: 4

Created about 4 years ago · Last pushed 8 months ago

Metadata Files

Readme Contributing License Citation

🦖 Rax: Learning-to-Rank using JAX

Rax is a Learning-to-Rank library written in JAX. Rax provides off-the-shelf implementations of ranking losses and metrics to be used with JAX. It provides the following functionality:

Ranking losses (rax.*_loss): rax.softmax_loss, rax.pairwise_logistic_loss, ...
Ranking metrics (rax.*_metric): rax.mrr_metric, rax.ndcg_metric, ...
Transformations (rax.*_t12n): rax.approx_t12n, rax.gumbel_t12n, ...

Ranking

A ranking problem is different from traditional classification/regression problems in that its objective is to optimize for the correctness of the relative order of a list of examples (e.g., documents) for a given context (e.g., a query). Rax provides support for ranking problems within the JAX ecosystem. It can be used in, but is not limited to, the following applications:

Search: ranking a list of documents with respect to a query.
Recommendation: ranking a list of items given a user as context.
Question Answering: finding the best answer from a list of candidates.
Dialogue System: finding the best response from a list of responses.

Synopsis

In a nutshell, given the scores and labels for a list of items, Rax can compute various ranking losses and metrics:

```python import jax.numpy as jnp import rax

scores = jnp.array([2.2, -1.3, 5.4]) # output of a model. labels = jnp.array([1.0, 0.0, 0.0]) # indicates doc 1 is relevant.

rax.ndcg_metric(scores, labels) # computes a ranking metric.

0.63092977

rax.pairwisehingeloss(scores, labels) # computes a ranking loss.

2.1

```

All of the Rax losses and metrics are purely functional and compose well with standard JAX transformations. Additionally, Rax provides ranking-specific transformations so you can build new ranking losses. An example is rax.approx_t12n, which can be used to transform any (non-differentiable) ranking metric into a differentiable loss. For example:

```python lossfn = rax.approxt12n(rax.ndcgmetric) lossfn(scores, labels) # differentiable approx ndcg loss.

-0.63282484

jax.grad(loss_fn)(scores, labels) # computes gradients w.r.t. scores.

[-0.01276882 0.00549765 0.00727116]

```

Installation

See https://github.com/jax-ml/jax#installation for instructions on installing JAX.

We suggest installing the latest stable version of Rax by running:

$ pip install rax

Examples

See the examples/ directory for complete examples on how to use Rax.

Citing Rax

If you use Rax, please consider citing our paper:

@inproceedings{jagerman2022rax, title = {Rax: Composable Learning-to-Rank using JAX}, author = {Rolf Jagerman and Xuanhui Wang and Honglei Zhuang and Zhen Qin and Michael Bendersky and Marc Najork}, year = {2022}, booktitle = {Proceedings of the 28th ACM SIGKDD International Conference on Knowledge Discovery \& Data Mining} }

Owner

Name: Google
Login: google
Kind: organization
Email: opensource@google.com
Location: United States of America

Website: https://opensource.google/
Twitter: GoogleOSS
Repositories: 2,773
Profile: https://github.com/google

Google ❤️ Open Source

Citation (CITATION.cff)

cff-version: 1.2.0
message: "If you use Rax, please consider citing our paper"
authors:
- family-names: "Jagerman"
  given-names: "Rolf"
  orcid: "https://orcid.org/0000-0002-5169-495X"
- family-names: "Wang"
  given-names: "Xuanhui"
- family-names: "Zhuang"
  given-names: "Honglei"
  orcid: "https://orcid.org/0000-0001-8134-1509"
- family-names: "Qin"
  given-names: "Zhen"
  orcid: "https://orcid.org/0000-0001-6739-134X"
- family-names: "Bendersky"
  given-names: "Michael"
- family-names: "Najork"
  given-names: "Marc"
  orcid: "https://orcid.org/0000-0003-1423-0854"
title: "Rax: Composable Learning-to-Rank using JAX"
url: "https://github.com/google/rax"
preferred-citation:
  type: conference-paper
  authors:
  - family-names: "Jagerman"
    given-names: "Rolf"
    orcid: "https://orcid.org/0000-0002-5169-495X"
  - family-names: "Wang"
    given-names: "Xuanhui"
  - family-names: "Zhuang"
    given-names: "Honglei"
    orcid: "https://orcid.org/0000-0001-8134-1509"
  - family-names: "Qin"
    given-names: "Zhen"
    orcid: "https://orcid.org/0000-0001-6739-134X"
  - family-names: "Bendersky"
    given-names: "Michael"
  - family-names: "Najork"
    given-names: "Marc"
    orcid: "https://orcid.org/0000-0003-1423-0854"
  collection-title: "Proceedings of the 28th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining"
  title: "Rax: Composable Learning-to-Rank using JAX"
  year: 2022

GitHub Events

Total

Release event: 1
Watch event: 14
Delete event: 9
Issue comment event: 2
Push event: 35
Pull request event: 18
Fork event: 1
Create event: 10

Last Year

Release event: 1
Watch event: 14
Delete event: 9
Issue comment event: 2
Push event: 35
Pull request event: 18
Fork event: 1
Create event: 10

Committers

Last synced: 7 months ago

All Time

Total Commits: 124
Total Committers: 14
Avg Commits per committer: 8.857
Development Distribution Score (DDS): 0.282

Past Year

Commits: 8
Committers: 6
Avg Commits per committer: 1.333
Development Distribution Score (DDS): 0.75

Top Committers

Name	Email	Commits
Rolf Jagerman	j**n@g**m	89
Xuanhui Wang	x**i@g**m	10
Jake VanderPlas	v**s@g**m	7
Honglei Zhuang	h**z@g**m	3
Rax Developers	r**v@g**m	3
Tianqi Liu	t**u@g**m	2
Peter Hawkins	p**s@g**m	2
Marcus Chiam	m**m@g**m	2
hemanthhari2000	h**0@g**m	1
Rebecca Chen	r**n@g**m	1
Neil Thomas	t**l@g**m	1
Jake Harmon	j**n@g**m	1
Daniel Suo	d**o@g**m	1
Nikitas Rontsis	n**r@g**m	1

Committer Domains (Top 20 + Academic)

google.com: 12

Issues and Pull Requests

Last synced: 7 months ago

All Time

Total issues: 3
Total pull requests: 152
Average time to close issues: 10 months
Average time to close pull requests: 5 days
Total issue authors: 3
Total pull request authors: 4
Average comments per issue: 0.33
Average comments per pull request: 0.08
Merged pull requests: 78
Bot issues: 1
Bot pull requests: 148

Past Year

Issues: 0
Pull requests: 24
Average time to close issues: N/A
Average time to close pull requests: 23 days
Issue authors: 0
Pull request authors: 1
Average comments per issue: 0
Average comments per pull request: 0.25
Merged pull requests: 12
Bot issues: 0
Bot pull requests: 24

View more stats

Top Authors

Issue Authors

allstar-app[bot] (1)
rjagerman (1)
travisbrady (1)

Pull Request Authors

copybara-service[bot] (148)
rjagerman (2)
Hemanthhari2000 (1)
thomas-a-neil (1)

Top Labels

Issue Labels

allstar (1) enhancement (1) wontfix (1)

Pull Request Labels

documentation (1)

Packages

Total packages: 1
Total downloads:
- pypi 27,485 last-month

Total dependent packages: 0
Total dependent repositories: 1
Total versions: 5
Total maintainers: 1

pypi.org: rax

Learning-to-Rank using JAX.

Homepage: https://github.com/google/rax
Documentation: https://rax.readthedocs.io/
License: Apache 2.0
Latest release: 0.4.0
published about 1 year ago

Versions: 5
Dependent Packages: 0
Dependent Repositories: 1
Downloads: 27,485 Last month

Rankings

Downloads: 3.4%

Stargazers count: 3.7%

Forks count: 8.9%

Average: 9.5%

Dependent packages count: 10.0%

Dependent repos count: 21.7%

Maintainers (1)

rax-dev

Last synced: 7 months ago

Dependencies

requirements/requirements-docs.txt pypi

myst-parser >=0.18
sphinx >=4
sphinx-autodoc-typehints >=1
sphinx_rtd_theme >=1
sphinxcontrib-bibtex >=2
sphinxcontrib-katex >=0.8

requirements/requirements-examples.txt pypi

clu >=0.0.6
flax >=0.3
optax >=0.1
tensorflow >=2
tensorflow-datasets >=4
tfds-nightly >=4.0.0.dev

requirements/requirements-test.txt pypi

absl-py >=1
numpy >=1.5.0
pytype >=2022

requirements/requirements.txt pypi

jax >=0.2.0
jaxlib >=0.1.0
typing_extensions >=3.7.0

.github/workflows/publish.yml actions

actions/checkout v2 composite
actions/setup-python v2 composite

.github/workflows/unittests.yml actions

actions/checkout v2 composite
actions/setup-python v2 composite

examples/t5x/requirements.txt pypi

seqio-nightly *

rax

Science Score: 44.0%

Keywords from Contributors

Repository

Basic Info

Statistics

Metadata Files

README.md

🦖 Rax: Learning-to-Rank using JAX

Ranking

Synopsis

0.63092977

2.1

-0.63282484

[-0.01276882 0.00549765 0.00727116]

Installation

Examples

Citing Rax

Owner

Citation (CITATION.cff)

GitHub Events

Total

Last Year

Committers

All Time

Past Year

Top Committers

Committer Domains (Top 20 + Academic)

Issues and Pull Requests

All Time

Past Year

Top Authors

Issue Authors

Pull Request Authors

Top Labels

Issue Labels

Pull Request Labels

Packages

pypi.org: rax

Rankings

Maintainers (1)

Dependencies