[Improve] Update benchmark scripts

[mzr1996]

This is a PR for mmcls 1.x

Motivation

Refactor and improve dev scripts

Modification

1. Refactor to support MMEngine.
2. Support --cfg-options to specify the same options

# AMP training benchmark
python .dev_scripts/benchmark_regression/3-benchmark_train.py mm_model \
    --job-name cls-amp --work-dir work_dirs/benchmark_train_amp \
    --cfg-options optim_wrapper.type='AmpOptimWrapper' optim_wrapper.loss_scale='dynamic'

3. Use --range to specify the training benchmark range

# Do quarterly training benchmark
python .dev_scripts/benchmark_regression/3-benchmark_train.py mm_model \
    --range quarter
# Do half-yearly training benchmark, but not include quarterly models
python .dev_scripts/benchmark_regression/3-benchmark_train.py mm_model \
    --range "=half-year"

4. Fix some training configs.
   • Use BCE in rsb-a1 training recipe
   • Fix the wrong ShuffleNet training schedule.

[codecov]

Codecov Report

Base: 0.02% // Head: 91.47% // Increases project coverage by +91.44% 🎉

Coverage data is based on head (dbcc7e8) compared to base (b8b31e9).
Patch coverage: 53.33% of modified lines in pull request are covered.

Additional details and impacted files

@@             Coverage Diff              @@
##           dev-1.x    #1028       +/-   ##
============================================
+ Coverage     0.02%   91.47%   +91.44%     
============================================
  Files          121      120        -1     
  Lines         8217     8230       +13     
  Branches      1368     1366        -2     
============================================
+ Hits             2     7528     +7526     
+ Misses        8215      571     -7644     
- Partials         0      131      +131     

Flag	Coverage Δ
unittests	91.47% <53.33%> (+91.44%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
mmcls/apis/inference.py	0.00% <0.00%> (ø)
mmcls/datasets/multi_label.py	100.00% <ø> (+100.00%)	⬆️
mmcls/models/backbones/timm_backbone.py	28.26% <0.00%> (+28.26%)	⬆️
mmcls/models/utils/attention.py	72.72% <10.44%> (+72.72%)	⬆️
mmcls/models/backbones/vision_transformer.py	74.31% <22.22%> (+74.31%)	⬆️
mmcls/datasets/dataset_wrappers.py	25.97% <22.53%> (+25.97%)	⬆️
mmcls/models/losses/label_smooth_loss.py	98.11% <83.33%> (+98.11%)	⬆️
mmcls/datasets/transforms/formatting.py	88.09% <95.65%> (+88.09%)	⬆️
mmcls/__init__.py	100.00% <100.00%> (+84.61%)	⬆️
mmcls/datasets/transforms/__init__.py	100.00% <100.00%> (+100.00%)	⬆️
... and 155 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[Ezra-Yu]

LGTM.