pytorch_pfn_extras.profiler.ChromeTracer#

class pytorch_pfn_extras.profiler.ChromeTracer(max_event_count=None, enable=True, append=True)#

Bases: Tracer

Tracer object that outputs a timeline in Chrome format.

Parameters:

  • max_event_count (int) – Limit the amount of events that can be traced, optional.

  • enable (bool) – Sets the tracer in active state. Optional, defaults to True.

  • append (bool) –

Methods

__init__([max_event_count, enable, append])

add_event(name)

add_remote_event(name, event)

clear()

enable(enable_flag)

finalize()

flush(filename, writer)

initialize_writer(filename, writer)

load_state_dict(to_load)

state_dict()

__init__(max_event_count=None, enable=True, append=True)#
Parameters:

  • max_event_count (Optional[int]) –

  • enable (bool) –

  • append (bool) –

Return type:

None

add_event(name)#
Parameters:

name (str) –

Return type:

Generator[None, None, None]

add_remote_event(name, event)#
Parameters:

  • name (str) –

  • event (Dict[str, Union[str, int, float]]) –

Return type:

None

clear()#
Return type:

None

enable(enable_flag)#
Parameters:

enable_flag (bool) –

Return type:

None

finalize()#
Return type:

None

flush(filename, writer)#
Parameters:

  • filename (str) –

  • writer (Writer) –

Return type:

None

initialize_writer(filename, writer)#
Parameters:

  • filename (str) –

  • writer (Writer) –

Return type:

None

load_state_dict(to_load)#
Parameters:

to_load (Dict[str, Any]) –

Return type:

None

state_dict()#
Return type:

Dict[str, Any]