For loop at line 325 in generate_perf_report_pytorch goes through dict_cat2names

[unagiboi]

There's a for loop that tries to create a dataframe for every operation category in dict_cat2names (which seems to be a pre-computed mapping), regardless of whether the operation categories are in the trace. The code doesn't fail, it just throws user warnings. But the user warnings could be confusing/alarming to the user.
I feel like there should be a simple if condition in the for loop where if the operation category isn't in the trace it continues.