Skip to main content

Project Configuration

The project configuration YAML file exists in the root of the project folder and holds the main parameters for the VAME workflow. The configuration file is created when initializing a new project with the init_new_project function.

The configuration file contains the following parameters:

Project parameters​

  • project_name (str): The name of the project.
  • creation_datetime (str): The creation datetime of the project.
  • model_name (str): The name of the model.
  • n_clusters (int): The number of clusters.
  • pose_confidence (float): The pose confidence.
  • project_path (str): The path to the project.
  • session_names (List[str]): The names of the sessions.
  • pose_estimation_filetype (PoseEstimationFiletype): The pose estimation filetype.

Data​

  • all_data (str): All data.
  • egocentric_data (bool): Egocentric data.
  • robust (bool): Robust data.
  • iqr_factor (int): IQR factor.
  • axis (str): Axis.
  • savgol_filter (bool): Savgol filter.
  • savgol_length (int): Savgol length.
  • savgol_order (int): Savgol order.
  • test_fraction (float): Test fraction.

RNN model general hyperparameters​

  • pretrained_model (str): Pretrained model.
  • pretrained_weights (bool): Pretrained weights.
  • num_features (int): Number of features.
  • keypoints (List[str]): Names of keypoints extracted from pose estimation data.
  • batch_size (int): Batch size.
  • max_epochs (int): Max epochs.
  • model_snapshot (int): Model snapshot.
  • model_convergence (int): Model convergence.
  • transition_function (str): Transition function.
  • beta (float): Beta.
  • beta_norm (bool): Beta normalization.
  • zdims (int): Zdims.
  • learning_rate (float): Learning rate.
  • time_window (int): Time window.
  • prediction_decoder (int): Prediction decoder.
  • prediction_steps (int): Prediction steps.
  • noise (bool): Noise.
  • scheduler (int): Scheduler.
  • scheduler_step_size (int): Scheduler step size.
  • scheduler_gamma (float): Scheduler gamma.
  • scheduler_threshold (float): Scheduler threshold.
  • softplus (bool): Softplus.

Segmentation​

  • segmentation_algorithms (List[SegmentationAlgorithms]): Segmentation algorithms.
  • hmm_trained (bool): HMM trained.
  • load_data (str): Load data.
  • individual_segmentation (bool): Individual segmentation.
  • random_state_kmeans (int): Random state kmeans.
  • n_init_kmeans (int): N init kmeans.

Video writer​

  • length_of_motif_video (int): Length of motif video.

UMAP parameter​

  • min_dist (float): Min dist.
  • n_neighbors (int): N neighbors.
  • random_state (int): Random state.
  • num_points (int): Num points.

RNN encoder hyperparameters​

  • hidden_size_layer_1 (int): Hidden size layer 1.
  • hidden_size_layer_2 (int): Hidden size layer 2.
  • dropout_encoder (float): Dropout encoder.

RNN reconstruction hyperparameters​

  • hidden_size_rec (int): Hidden size rec.
  • dropout_rec (float): Dropout rec.
  • n_layers (int): N layers.

RNN prediction hyperparameters​

  • hidden_size_pred (int): Hidden size pred.
  • dropout_pred (float): Dropout pred.

RNN loss hyperparameters​

  • mse_reconstruction_reduction (str): MSE reconstruction reduction.
  • mse_prediction_reduction (str): MSE prediction reduction.
  • kmeans_loss (int): Kmeans loss.
  • kmeans_lambda (float): Kmeans lambda.
  • anneal_function (str): Anneal function.
  • kl_start (int): KL start.
  • annealtime (int): Annealtime.