To ease the reproduction of papers, and the upload of full models on HuggingFace, the xpmir.papers package can be used.


Papers experimental parameters are defined by data classes. The main one is xpmir.papers.helpers.PaperExperiment that defines an id (for experimaestro), a title and a description. These informations can be used e.g. when uploading the trained models on HuggingFace.


alias of ConfigurationBase

class xpmir.papers.helpers.NeuralIRExperiment(*, id: str = '???', file: str = 'experiment', parent: str | None = None, title: str = '', subtitle: str = '', paper: str = '', description: str = '', gpu: bool = True, seed: int = 0)[source]

Bases: ConfigurationBase

Settings most neural IR experiments

description: str

Description of the experiment

file: str

Relative path of the file containing a run function

gpu: bool

Use GPU for computation

id: str

ID of the experiment

This ID is used by experimaestro when running as the experiment.

paper: str

Source paper for this experiment

parent: str | None

Relative path of a YAML file that should be merged

seed: int

The seed used for experiments

subtitle: str

Allows to give some more details about the experiment

title: str

Short description of the experiment


Pipelines factorize the code necessary to run some experiments: for instance, training re-rankers on MS-Marco is usually performed with similar training data, evaluation datasets.

Implemented papers

This page give some information about reproduction of papers based on XPMIR.