FRAME_FM package

Subpackages

• FRAME_FM.dataloaders package
  • Submodules
  • FRAME_FM.dataloaders.combined_dataloader module
    • CombinedDataModule
  • FRAME_FM.dataloaders.demo_eurosat module
    • EuroSATDataModule
      • EuroSATDataModule.prepare_data()
  • Module contents
• FRAME_FM.datasets package
  • Class Hierarchy
  • Changing datasets using preprocessors and transforms
  • Submodules
  • FRAME_FM.datasets.ImageLabel_Dataset module
    • TransformedDataset
  • FRAME_FM.datasets.InputOnly_Dataset module
    • TransformedInputCoordsDataset
    • TransformedInputDataset
  • Module contents
• FRAME_FM.models package
  • Submodules
  • FRAME_FM.models.demo_autoencoder module
    • EuroSATAutoencoder
      • EuroSATAutoencoder.configure_optimizers()
      • EuroSATAutoencoder.forward()
      • EuroSATAutoencoder.test_step_body()
      • EuroSATAutoencoder.training_step_body()
      • EuroSATAutoencoder.validation_step_body()
  • FRAME_FM.models.convae module
    • ConvAutoencoder
      • ConvAutoencoder.configure_optimizers()
      • ConvAutoencoder.forward()
      • ConvAutoencoder.on_validation_epoch_end()
      • ConvAutoencoder.on_validation_epoch_start()
      • ConvAutoencoder.training_step_body()
      • ConvAutoencoder.validation_step_body()
  • FRAME_FM.models.mmmae module
    • MultimodalMaskedAutoencoder
      • MultimodalMaskedAutoencoder.configure_optimizers()
      • MultimodalMaskedAutoencoder.forward()
      • MultimodalMaskedAutoencoder.forward_decoder()
      • MultimodalMaskedAutoencoder.forward_encoder()
      • MultimodalMaskedAutoencoder.forward_loss()
      • MultimodalMaskedAutoencoder.initialize_weights()
      • MultimodalMaskedAutoencoder.input_embedders
      • MultimodalMaskedAutoencoder.random_masking()
      • MultimodalMaskedAutoencoder.test_step_body()
      • MultimodalMaskedAutoencoder.training_step_body()
      • MultimodalMaskedAutoencoder.validation_step_body()
  • Module contents
• FRAME_FM.training package
  • Submodules
  • FRAME_FM.training.logging_utils module
    • create_csv_logger()
    • create_mlflow_logger()
  • FRAME_FM.training.train module
    • main()
  • Module contents
• FRAME_FM.transforms package
  • Module contents
• FRAME_FM.utils package
  • Submodules
  • FRAME_FM.utils.LightningDataModuleWrapper module
    • BaseDataModule
      • BaseDataModule.setup()
      • BaseDataModule.test_dataloader()
      • BaseDataModule.train_dataloader()
      • BaseDataModule.val_dataloader()
  • FRAME_FM.utils.LightningModuleWrapper module
    • BaseModule
      • BaseModule.log_metrics()
      • BaseModule.test_step()
      • BaseModule.test_step_body()
      • BaseModule.training_step()
      • BaseModule.training_step_body()
      • BaseModule.validation_step()
      • BaseModule.validation_step_body()
  • FRAME_FM.utils.embedders module
    • BaseEmbedder
      • BaseEmbedder.embed_dim
      • BaseEmbedder.forward()
      • BaseEmbedder.initialize_weights()
      • BaseEmbedder.n_patches
      • BaseEmbedder.reconstruct_dim
      • BaseEmbedder.reconstruct_tokens()
      • BaseEmbedder.tokenify()
      • BaseEmbedder.untokenify()
    • BoundedPatchEmbed
    • PatchEmbed
      • PatchEmbed.forward()
      • PatchEmbed.initialize_weights()
      • PatchEmbed.reconstruct_tokens()
      • PatchEmbed.tokenify()
      • PatchEmbed.untokenify()
    • STPatchEmbed
      • STPatchEmbed.forward()
      • STPatchEmbed.initialize_weights()
      • STPatchEmbed.tokenify()
    • calc_embed_omega()
    • partition_embed_dim()
    • sincos_embed_coords()
  • Module contents

Submodules

FRAME_FM.cli module

Click entrypoint.

FRAME_FM.cli.check_configs_directory()

Check if the configs directory exists, and warn the user if not.

FRAME_FM.cli.check_torchx_config()

Ensure .torchxconfig exists to avoid TorchX initialization errors.

FRAME_FM.cli.display_contents_of_config_file(torchx_only: bool, config_file: str) → None

Display the contents of a config file.

Parameters:

• torchx_only – If True, only display the contents of the torchx config.

• config_file – The file to search for within the config directory.

FRAME_FM.cli.edit_config_file(config_file: str, key_value_pairs: str) → None

Edit the config file, raising errors if the format is incorrect, or the key is missing.

FRAME_FM.cli.edit_torch_config_file(key_value_pairs: str) → None

Edit the key-value pairs within the torchxconfig TOML.

Parameters:

key_value_pairs – String representations of keys and their new values to be edited.

FRAME_FM.cli.get_hydra_cfg(overrides: tuple[str, ...])

Safely compose Hydra config, clearing any existing global state first.

FRAME_FM.cli.get_torchx_config() → dict

Reads .torchxconfig manually to avoid import errors.

FRAME_FM.cli.launch_torchx_job(cfg, platform_cfg, scheduler: str, overrides: tuple[str, ...])

Dispatches the command to TorchX.

FRAME_FM.cli.show_config_files(torchx_only: bool) → None

Output a structured list of configuration folders and their YAML contents.

Parameters:

torchx_only – If True, only locate and show the full path to the torchx config.

FRAME_FM.cli.train_run_with_local_hydra(verbose: bool, overrides: tuple[str, ...]) → None

FRAME_FM.cli.view_hydra_defaults() → None

Display the Hydra default values from the config.

Module contents