scripts folder

This folder contains scripts to run various steps of the pipeline from the command line or via snakemake.

Subfolders

• scripts/evaluation folder
  • plot_training_metrics.py script
  • plot_embedding_edge_performance.py script
  • plot_embedding_best_tracking_performance.py script
  • plot_embedding_edge_plane_diff.py script
  • plot_embedding_n_neigbhours.py script
  • plot_gnn_edge_performance.py script
  • plot_gnn_triplet_performance.py script
  • plot_gnn_best_tracking_performance.py script
  • evaluate_allen_on_test_sample.py script
    • evaluate_allen()
    • get_event_ids()
  • evaluate_etx4velo.py script
    • compute_n_shared_hits()
    • compute_plane_stats()
    • evaluate()
    • evaluate_partition()
    • perform_evaluation()
  • compare_allen_vs_etx4velo.py script
    • compare_allen_vs_etx4velo_from_trackevaluators()
    • compare_etx4velo_vs_allen()
    • compare_etx4velo_vs_allen_global()
• scripts/plotfactory folder
  • plot_data_statistics.py script
  • plot_embedding_explanation.py script
    • filter_edges_planes()
    • get_df_edges()
    • get_fig_ax()
    • get_random_batch_file()
    • plot_edges()
    • plot_hits()
  • plot_plane_edge_statistics.py script
    • get_df_hits_particles()
• scripts/trackfactory folder
  • trackfactory.analyse_long_electron_clones.py script
  • trackfactory/plot_crossing_tracks.py script
  • trackfactory/plot_shared_electron_tracks.py script
  • trackfactory.dfutils module
    • add_particle_information()
    • compute_edge_counts()
    • compute_n_particles_hit()
    • find_connected_particle_ids()
    • get_df_edges()
    • get_df_hits_particles_from_batch()
    • no_shared_edges()
    • plot_connected_particle_ids()

collect_test_samples.py script

A script that generates the test_samples.yaml file that defines the test samples used in this repository.

preprocess_test_sample.py script

A script that runs the preprocessing of a test sample.

process_test_sample.py script

A script that runs the preprocessing of a test sample.

build_graph_using_embedding.py script

A script that runs the graph building using the embedding model learnt at the previous stage.

scripts.build_graph_using_embedding.run(path_or_config, partitions=['train', 'val', 'test'], checkpoint=None, reproduce=True, use_gpu=True, suffix=None, **kwargs)

Run the inference of the metric learning stage.

Parameters:

• path_or_config (str | dict) – configuration dictionary, or path to the YAML file that contains the configuration

• partitions (List[str]) –

  Partitions to run the inference on:

  • train: train dataset

  • val: validation dataset

  • test: all the test datasets

  • A specific test dataset name

• checkpoint (UnionType[EmbeddingBase, str, None]) – Model already loaded, or path to its checkpoint. If None, try to find it automatically in the artifact folder given the configuration.

• reproduce (bool) – whether to delete an existing folder

• use_gpu (bool) – whether to use the GPU (if available)

• **kwargs – Other keyword arguments passed to the PyTorch.LightingModel.load_from_checkpoint() class method

train_model.py script

A script that runs the training of a model (embedding or GNN).

scripts.train_model.get_parsed_args()

Return type:

Namespace

scripts.train_model.train_model(path_or_config, step, identifier=None)

Run the training of a model.

Parameters:

• path_or_config (str | dict) – pipeline configuration or path to it.

• step (str) – Model step, such as embedding or gnn.

• identifier (Optional[str]) – Identifier added at the end of the step name.

Return type:

Tuple[Trainer, Module]

Returns:

Trainer and trained model.

build_tracks.py script

Script that runs the edge filtering, triplet building and filtering, and track building from triplets.

scripts.build_tracks.build(path_or_config, partitions=['train', 'val', 'test'], checkpoint=None, reproduce=True, edge_score_cut=None, triplet_score_cut=None, single_edge_score_cut=None, strategy=None, with_triplets=True, **kwargs)

export_model_to_onnx.py script

A python script to export a model to an ONNX file.

scripts.export_model_to_onnx.export_model_to_onnx(path_or_config, step, mode=None, output_path=None, options=None, dummy=False)

Export a model of a pipeline to an ONNX file.

Parameters:

• path_or_config (str | dict) – Path to the pipeline configuration file or the configuration dictionary.

• step (Literal['embedding', 'gnn']) – Model step, such as embedding or gnn.

• mode (Optional[str]) – Export mode.

• output_path (Optional[str]) – Path where to save the .onnx file containing the model. If not provided, it is defined from the experiment name and step.

• **options (Optional[Iterable[str]]) – export options

Return type:

None

scripts.export_model_to_onnx.get_parsed_args()

Return type:

Namespace