olympus.tasks.task module

class olympus.tasks.task.GenerateSummary

Bases: object

Methods

get_name
is_nested
print
rename
retrieve_nested
task_summary

dispatch = {'DataLoader': <function GenerateSummary.<lambda> at 0x7f737cbd90e0>, 'LRSchedule': <function GenerateSummary.<lambda> at 0x7f737cbd9290>, 'MetricList': <function GenerateSummary.<lambda> at 0x7f737cbd9200>, 'Model': <function GenerateSummary.<lambda> at 0x7f737cbeff80>, 'Optimizer': <function GenerateSummary.<lambda> at 0x7f737cbd9050>, 'TransformedSubset': <function GenerateSummary.<lambda> at 0x7f737cbd9170>}

get_name(attr, obj, type_name, depth=0)

is_nested(name)

print(msg='', end='\n')

rename(name)

retrieve_nested(name, obj)

task_summary(obj)

class olympus.tasks.task.Task(device=None)

Bases: object

Attributes:

device

events

metrics

Methods

eval_loss(batch)	This is used to compute validation and test loss
fit(epoch[, context])	Execute a single batch
get_space()	Return missing hyper parameters that need to be set using init
init(**kwargs)	Used to initialize the hyperparameters is any
load_state_dict(state[, strict])	Try to load a previous unfinished state to resume
state_dict([destination, prefix, keep_vars])	Save a state the task can go back to if an error occur

report
resumed
set_device
summary

device

eval_loss(batch)

This is used to compute validation and test loss

events

fit(epoch, context=None)

Execute a single batch

Parameters:

epoch: int

current step in the training process

context: dict

Optional Context

Notes

You should wrap whatever code you have here inside a BadResumeGuard to prevent users from resuming a failed task that can have a bad states

To resume a task, you need to create a clean one with the same hyper parameters. It will pickup automatically where at its last checkpoint

get_space()

Return missing hyper parameters that need to be set using init

init(**kwargs)

Used to initialize the hyperparameters is any

load_state_dict(state, strict=True)

Try to load a previous unfinished state to resume

Notes

You should wrap whatever code you have here inside a BadResumeGuard to prevent users from resuming a failed task that can have a bad states

To resume a task, you need to create a clean one with the same hyper parameters. It will pickup automatically where at its last checkpoint

metrics

report(pprint=True, print_fun=<built-in function print>)

resumed()

set_device(device)

state_dict(destination=None, prefix='', keep_vars=False)

Save a state the task can go back to if an error occur

summary()