autointent.context.Context#
- class autointent.context.Context(seed=42)#
Context manager for configuring and managing data handling, vector indexing, and optimization.
This class provides methods to set up logging, configure data and vector index components, manage datasets, and retrieve various configurations for inference and optimization. Not intended to be instantiated by user.
- Parameters:
seed (int | None)
- data_handler: autointent.context.data_handler.DataHandler#
Convenient wrapper for
autointent.Dataset
.
- optimization_info: autointent.context.optimization_info.OptimizationInfo#
Object for storing optimization trials and inter-node communication.
- callback_handler#
Internal callback for logging to tensorboard or wandb.
- seed = 42#
- configure_logging(config)#
Configure logging settings.
- Parameters:
config (autointent.configs.LoggingConfig) – Logging configuration settings.
- Return type:
None
- configure_transformer(config)#
Configure the vector index client and embedder.
- Parameters:
config (autointent.configs.EmbedderConfig | autointent.configs.CrossEncoderConfig) – configuration for the transformers to use during optimization.
- Return type:
None
- set_dataset(dataset, config)#
Set the datasets for training, validation and testing.
- Parameters:
dataset (autointent.Dataset) – dataset to use during optimization.
config (autointent.configs.DataConfig) – data configuration settings.
- Return type:
None
- dump()#
Save all information about optimization process to disk.
Save metrics, hyperparameters, inference, configurations, and datasets to disk.
- Return type:
None
- get_dump_dir()#
Get the directory for saving dumped modules.
Return path to the dump directory or None if dumping is disabled.
- Return type:
pathlib.Path | None
- is_multilabel()#
Check if the dataset is configured for multilabel classification.
- Return type:
- resolve_embedder()#
Resolve the embedder configuration.
Returns the best embedder configuration or default configuration.
- Raises:
RuntimeError – If embedder configuration cannot be resolved.
- Return type:
- resolve_ranker()#
Resolve the cross-encoder configuration.
Returns default config if set.
- Raises:
RuntimeError – If cross-encoder configuration cannot be resolved.
- Return type: