Closed mghosh00 closed 6 months ago
Allow extra flexibility within the model by allowing continuous labels rather than just categorical ones. This will involve changing/adding the following classes:
components
Network.__init__
Network.forward_pass_one_datapoint
else
Network.store_gradient_of_loss
tests
main
AbstractSimulator.__init__
AbstractSimulator.forward_pass_one_batch
AbstractSimulator.abs_generate_scatter
Plotter
Trainer.run
Tester.run
Validator.validate
_update_categorical_dataframe
Plotter.comparison_scatter
functions
Loss
CrossEntropyLoss
MSELoss
Allow extra flexibility within the model by allowing continuous labels rather than just categorical ones. This will involve changing/adding the following classes:
components
Network.__init__
Network.forward_pass_one_datapoint
(else
block)Network.store_gradient_of_loss
(output layer)tests
main
AbstractSimulator.__init__
AbstractSimulator.forward_pass_one_batch
AbstractSimulator.abs_generate_scatter
(change the method whichPlotter
calls)Trainer.run
,Tester.run
,Validator.validate
(just don't call_update_categorical_dataframe
)Plotter.comparison_scatter
(plot against line y = x)tests
functions
Loss
(rename toCrossEntropyLoss
)MSELoss
(remember to just return the squared difference)tests