[BUG REPORT] total memory_wirte and total memory_read wrong.

[lllukehuang]

In the total result row, the results of total memory_wirte and total memory_read are incorrect, which just show the last layer's result.

torchstat/torchstat/reporter.py

Lines 57 to 62 in b52a3b0

	total_df = pd.Series([total_parameters_quantity, total_memory,
	total_operation_quantity, total_flops,
	total_duration, mread, mwrite, total_memrw],
	index=['params', 'memory(MB)', 'MAdd', 'Flops', 'duration[%]',
	'MemRead(B)', 'MemWrite(B)', 'MemR+W(B)'],
	name='total')

Here should be total_mread and 'total_mwrite' rather than mread and 'mwrite'.