graph LR
PerforatedAI_Training["PerforatedAI.Training"]
PerforatedAI_Evaluation["PerforatedAI.Evaluation"]
PerforatedAI_Data["PerforatedAI.Data"]
PerforatedAI_Models["PerforatedAI.Models"]
PerforatedAI_Metrics["PerforatedAI.Metrics"]
PerforatedAI_Config["PerforatedAI.Config"]
PerforatedAI_Utils["PerforatedAI.Utils"]
PerforatedAI_API["PerforatedAI.API"]
PerforatedAI_Training -- "consumes data from" --> PerforatedAI_Data
PerforatedAI_Training -- "trains models defined in" --> PerforatedAI_Models
PerforatedAI_Evaluation -- "uses to compute scores" --> PerforatedAI_Metrics
PerforatedAI_Evaluation -- "utilizes for support tasks" --> PerforatedAI_Utils
PerforatedAI_Training -- "reports progress to" --> PerforatedAI_Metrics
PerforatedAI_Training -- "retrieves hyperparameters from" --> PerforatedAI_Config
PerforatedAI_Training -- "utilizes for support tasks" --> PerforatedAI_Utils
PerforatedAI_API -- "offers high-level training functions to" --> PerforatedAI_Training
One paragraph explaining the functionality which is represented by this graph. What the main flow is and what is its purpose.
Manages optimizers, loss functions, and the core training loops. It is responsible for iterating through data, performing forward and backward passes, and updating model parameters.
Related Classes/Methods: None
Provides tools and routines for calculating performance metrics, validating models, and generating reports based on trained models and test datasets.
Related Classes/Methods: None
Handles data loading, preprocessing, augmentation, and batching, providing structured input to the training and evaluation processes.
Related Classes/Methods: None
Defines and manages neural network architectures, including pre-built models, custom model definitions, and model serialization/deserialization.
Related Classes/Methods:
perforatedai.pb_models
Provides a collection of performance metrics (e.g., accuracy, loss, precision, recall) and mechanisms for tracking and reporting them during training and evaluation.
Related Classes/Methods: None
Manages configuration parameters, hyperparameters, and experiment settings, allowing for reproducible and flexible experimentation.
Related Classes/Methods: None
Contains common utility functions, helper methods, and general-purpose tools used across various components of the library, promoting code reusability.
Related Classes/Methods:
perforatedai.pb_utils
Exposes high-level, user-friendly interfaces and functions for initiating training runs, evaluating models, and interacting with the core functionalities of the library.
Related Classes/Methods: None