Quick start

Installation

Install with pip:

pip install nrslib

Main Technologies

Familarity with the follow tools are recommended to use this library.

PyTorch Lightning - a lightweight PyTorch wrapper for high-performance AI research. Think of it as a framework for organizing your PyTorch code.

Hydra - a framework for elegantly configuring complex applications. The key feature is the ability to dynamically create a hierarchical configuration by composition and override it through config files and the command line.

TorchMetrics - a collection of 80+ PyTorch metrics implementations and an easy-to-use API to create custom metrics.

Lightning-Hydra-Template - A clean and scalable template to kickstart your deep learning project.

Running a model

Import the library:

import nrslib

Export default configs:

nrslib.export_default_config(“~/configs”)

Choose and edit configs in the open console dialog and start training afterwards:

nrslib.start_train(“~/configs”, [“experiment=naml.yaml”])

To test a model, run:

nrslib.start_test(“~/configs”, [“experiment=naml.yaml”])

To extend a model or datamodule inherit it:

from nrslib.src.models.naml import NAML
class ImprovedNAML(NAML):
    . . . .

Then adjust the configurations to use the new class:

_target_: path.to.class.ImprovedNAML

Using Torchmetrics

Metrics that should be used for model evaluation and testing need to be added to the models .yaml configuration file. Example: The .yaml file for the Neural Collaborative Filtering model could look as follows:

_target_: src.models.collaborativeFiltering.CollaborativeFiltering
lr: 0.001
n_factors: 100
data_dir: ${data_dir}/MIND
metrics:
  valid:
    accuracy:
      _target_: torchmetrics.Accuracy
    ndcg5:
      _target_: torchmetrics.RetrievalNormalizedDCG
      k: 5
  test:
    accuracy:
      _target_: torchmetrics.Accuracy
    ndcg5:
      _target_: torchmetrics.RetrievalNormalizedDCG
      k: 5