pytorch / benchmark
Features of Interest
Aspects in the source code identified through RegEx patterns.
Intro
Features of interest are any aspects of a software system that can be identified through patterns in code.
Features of interest provide you with a way to focus your attention on relevant parts of the codebase.
Typical examples include, security, TODOs, logging.
A feature of interest may be present in multiple files. Any source code file may be in zero or multiple features of interest.
Overview
GENERAL
TODOs
2,633 LOC (4%)
16 files
|
found text
|
found text per file
1 GENERAL Concerns
The "general" group contains
2
concerns.
TODOs
Unclassified
general
1.1 TODOs
2,633 LOC (4%)
16 files
|
found text
|
found text per file
The following criteria are used to filter files:
files with any line of content like "
.*(TODO|FIXME)( |:| ).*
".
16
files match defined criteria (
2,633
lines of code,
4.7%
vs. main code). All matches are in *.py files.
33
lines match the content pattern.
details...
per component - primary logical decomposition
torchbenchmark/models/tacotron2
in 1 file 92 LOC (2%)
torchbenchmark/models/demucs
in 1 file 238 LOC (10%)
ROOT
in 1 file 139 LOC (8%)
torchbenchmark/models/attention_is_all_you_need_pytorch
in 4 files, 517 LOC (37%)
torchbenchmark/models/speech_transformer
in 1 file 95 LOC (8%)
torchbenchmark/util/torchtext_legacy
in 2 files, 227 LOC (22%)
torchbenchmark/models/moco
in 1 file 357 LOC (35%)
torchbenchmark/score
in 1 file 105 LOC (12%)
torchbenchmark/models/tts_angular
in 1 file 186 LOC (39%)
torchbenchmark
in 1 file 390 LOC (87%)
torchbenchmark/e2e_models/hf_bert
in 1 file 167 LOC (54%)
torchbenchmark/microbenchmarks/nvfuser
in 1 file 120 LOC (96%)