torchcell.trainers

PyTorch Lightning training tasks. The package exports the regression task (RegressionTask), a simple linear baseline, and two DCell tasks. The fit_int_* and int_* modules hold the tasks used by individual experiments and are imported by module path from the experiment scripts. Every class below is listed under the module that defines it.

`dcell_regression`

Lightning training task for DCell graph-based fitness regression.

Classes 

dcell_regression.DCellRegressionTask

LightningModule for training models on graph-based regression datasets.

`dcell_regression_slim`

Slim Lightning trainer for DCell regression with subsystem and root metrics.

Classes 

dcell_regression_slim.DCellRegressionSlimTask

LightningModule for training models on graph-based regression datasets.

`int_dango`

Lightning training task for the DANGO genetic-interaction regression model.

Classes 

int_dango.RegressionTask

Lightning module training DANGO to regress genetic interaction scores.

`int_dcell`

Lightning training module for DCell gene-interaction regression.

Classes 

int_dcell.RegressionTask

Lightning module training a DCell model to regress gene interactions.

`int_hetero_cell`

Lightning regression tasks for heterogeneous-cell gene interaction models.

Classes 

`int_hetero_cell.RegressionTask`	Lightning task training a model to predict gene interaction scores.
`int_hetero_cell.DiffusionRegressionTask`	`RegressionTask` for a diffusion decoder (006 `hetero_cell_bipartite_dango_diff_gi`).

Functions 

`int_hetero_cell.scheduler_type`	Return the scheduler `type` of an lr scheduler config, refusing a missing or unknown one by name (a misspelled type used to become a plateau scheduler).
`int_hetero_cell.normalize_accumulation_schedule`	Return the schedule with integer epoch keys in ascending order.
`int_hetero_cell.accumulation_steps`	Steps of the last threshold reached (`epoch >= threshold`); 1 before any.

`int_transformer_cell`

Lightning training module for the transformer cell gene-interaction model.

Classes 

int_transformer_cell.RegressionTask

Lightning module training the transformer cell model on gene interactions.

`neo_regression`

Lightning regression trainer with MSE/ListMLE losses and correlation metrics.

Classes 

`neo_regression.ListMLEMetric`	TorchMetric tracking the running mean ListMLE loss over batches.
`neo_regression.MSEListMLELoss`	Loss combining MSE with an alpha-weighted ListMLE ranking term.
`neo_regression.RegressionTask`	LightningModule for training models on graph-based regression datasets.

`simple_linear_regression`

Lightning task for simple linear regression on graph perturbation data.

Classes 

simple_linear_regression.SimpleLinearRegressionTask

LightningModule for training models on graph-based regression datasets.