XGBoost: Difference between revisions
mNo edit summary |
No edit summary |
||
| (9 intermediate revisions by 5 users not shown) | |||
| Line 1: | Line 1: | ||
<languages /> | |||
[[Category:Software]][[Category:AI and Machine Learning]] | |||
<translate> | <translate> | ||
<!--T:1--> | |||
'''[https://xgboost.readthedocs.io/en/latest/ XGBoost]''' ''is an optimized distributed gradient boosting library designed to be highly efficient, flexible and portable''. It is a popular package used for a wide variety of machine learning and datascience tasks, serving the role of a convenient, domain-agnostic black box classifier. XGBoost provides GPU accelerated learning for some problems, and Compute Canada provides a GPU enabled build. | '''[https://xgboost.readthedocs.io/en/latest/ XGBoost]''' ''is an optimized distributed gradient boosting library designed to be highly efficient, flexible and portable''. It is a popular package used for a wide variety of machine learning and datascience tasks, serving the role of a convenient, domain-agnostic black box classifier. XGBoost provides GPU accelerated learning for some problems, and Compute Canada provides a GPU enabled build. | ||
== Python Module == | <!--T:2--> | ||
A very common way to use XGBoost is though its python interface, provided as the <code>xgboost</code> python module. Compute Canada provides an optimized, multi-GPU enabled build as a [[Python]] wheel | For detailed documentation on using the library, please consult the [https://xgboost.readthedocs.io/en/latest/get_started.html xgboost documentation]. There is a [https://xgboost.readthedocs.io/en/latest/gpu/index.html separate section for GPU-enabled training]. | ||
== Python Module Installation == <!--T:3--> | |||
A very common way to use XGBoost is though its python interface, provided as the <code>xgboost</code> python module. Compute Canada provides an optimized, multi-GPU enabled build as a [[Python]] wheel; readers can should familiarize themselves with the use of [[Python#Creating_and_using_a_virtual_environment | Python virtual environments]] before starting an XGBoost project. | |||
<!--T:4--> | |||
Currently, version 0.81 of XGBoost is available. The following commands illustrate the needed package and module: | Currently, version 0.81 of XGBoost is available. The following commands illustrate the needed package and module: | ||
{{Commands| | {{Commands | ||
prompt=(myvenv) name@server $ | |prompt=(myvenv) name@server $ | ||
|module load nixpkgs/16.09 intel/2018.3 cuda/10.0.130 | |module load nixpkgs/16.09 intel/2018.3 cuda/10.0.130 | ||
|module load nccl/2.3.5 | |module load nccl/2.3.5 | ||
|pip install xgboost==0.81 --no-index | |pip install xgboost{{=}}{{=}}0.81 --no-index | ||
}} | }} | ||
</translate> | </translate> | ||
Latest revision as of 19:19, 31 May 2024
XGBoost is an optimized distributed gradient boosting library designed to be highly efficient, flexible and portable. It is a popular package used for a wide variety of machine learning and datascience tasks, serving the role of a convenient, domain-agnostic black box classifier. XGBoost provides GPU accelerated learning for some problems, and Compute Canada provides a GPU enabled build.
For detailed documentation on using the library, please consult the xgboost documentation. There is a separate section for GPU-enabled training.
Python Module Installation
A very common way to use XGBoost is though its python interface, provided as the xgboost python module. Compute Canada provides an optimized, multi-GPU enabled build as a Python wheel; readers can should familiarize themselves with the use of Python virtual environments before starting an XGBoost project.
Currently, version 0.81 of XGBoost is available. The following commands illustrate the needed package and module:
(myvenv) name@server $ module load nixpkgs/16.09 intel/2018.3 cuda/10.0.130
(myvenv) name@server $ module load nccl/2.3.5
(myvenv) name@server $ pip install xgboost==0.81 --no-index