[REQUEST] Support profiling backward using Flops Profiler as a standalone package

[yzs981130]

Is your feature request related to a problem? Please describe.
I try to use flops profiler as a standalone package to profiling the backward of a custom model. I have followed both two instructions, here and here. However, both examples only consider profiling the forward stage. I check the code and it seems that it currently only prints the bwd_latency when using runtime as below

DeepSpeed/deepspeed/profiling/flops_profiler/profiler.py

Lines 313 to 317 in 5e42cc8

	if self.ds_engine and self.ds_engine.wall_clock_breakdown():
	bwd_latency = self.ds_engine.timers('backward').elapsed(False) / 1000.0
	step_latency = self.ds_engine.timers('step').elapsed(False) / 1000.0
	print('{:<60} {:<8}'.format('bwd latency: ',
	duration_to_string(bwd_latency)))

Describe the solution you'd like
Is there any approach to profiling the backward by the flops profiler?