trivialaugment:这是TrivialAugment的正式实现，也是应用多种增强策略的迷你图书馆
需积分: 43 194 浏览量
2021-03-25
15:31:46
上传
评论
收藏 120KB ZIP 举报
trivialaugment:这是TrivialAugment的正式实现，也是应用多种增强策略的迷你图书馆 （100个子文件） 
.gitignore 1KB LICENSE 1KB README.md 3KB fast_autoaugment.py 113KB autoaugment.py 21KB aug_lib.py 19KB train.py 19KB data.py 13KB wideresnet.py 9KB resnet.py 6KB common.py 3KB __init__.py 3KB create_variants_of_set_config.py 3KB imagenet.py 3KB shake_resnext.py 3KB aggregate_results.py 3KB shake_resnet.py 3KB augmentations.py 2KB metrics.py 2KB convnet.py 2KB shakeshake.py 1KB lr_scheduler.py 661B mlp.py 616B evaluation_tools.py 282B __init__.py 0B __init__.py 0B
requirements.txt 252B resnet50_imagenet_270epochs_4x8xb64_fixedlr_ta_fixed.yaml 392B resnet50_smallwidthimagenet_270epochs_4x8xb64_fixedlr_ta_fixed.yaml 381B shake26_2x96d_cifar10_b128_.2lr_8times_nowarmup_ta_fix_600epochs.yaml 348B shake26_2x96d_cifar10_b128_.2lr_8times_nowarmup_ta_600epochs.yaml 347B wresnet28x10_svhncore_b128_maxlr.1_ta_fixedsearchspace_2.yaml 346B wresnet28x10_svhncore_b128_maxlr.1_ta_fixedsearchspace.yaml 344B wresnet28x10_cifar100_b128_maxlr.1_weighteduniaug_w=0,1_fixedsesp_nowarmup_8timesbatch_200epochs.yaml 343B wresnet28x10_svhncore_b128_maxlr.1_ta_paperautoaugsearchspace_2_nowarmup_fixedpreprocessing.yaml 343B wresnet28x10_svhncore_b128_maxlr.1_ta_fixpaperautoaugsearchspace_2_nowarmup_fixedpreprocessing.yaml 343B wresnet28x10_cifar100_b128_maxlr.1_weighteduniaug_w=0,1_widesesp_nowarmup_8timesbatch_200epochs.yaml 342B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_widesesp_nowarmup_8timesbatch_200epochs.yaml 341B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_autoaugpapersesp_nowarmup_200epochs.yaml 340B wresnet28x10_svhn_b128_maxlr.1_ta_fixedsearchspace.yaml 339B wresnet28x10_svhncore_b128_maxlr.1_ta_fixedsearchspace_threestrengths_2_nowarmup_fixedpreprocessing.yaml 335B wresnet28x10_svhncore_b128_maxlr.1_ta_fixedsearchspace_twostrengths_2_nowarmup_fixedpreprocessing.yaml 335B wresnet28x10_cifar100_b128_maxlr.1_weighteduniaug_w=0,1_3strengths_fixedsesp_nowarmup_200epochs.yaml 334B wresnet28x10_cifar100_b128_maxlr.1_weighteduniaug_w=0,1_2strengths_fixedsesp_nowarmup_200epochs.yaml 334B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_2strengths_fixedsesp_nowarmup_200epochs.yaml 333B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_3strengths_fixedsesp_nowarmup_200epochs.yaml 333B wresnet28x10_cifar100_b128_maxlr.1_randaug_fixedsesp_200epochs_2.yaml 326B wresnet28x10_cifar100_b128_maxlr.1_randaug_fixedsesp_200epochs.yaml 326B wresnet28x10_svhn_b128_maxlr.1_ta_fixedsearchspace_2_nowarmup.yaml 324B wresnet28x10_svhncore_b128_maxlr.1_ta_ohlsearchspace_2_nowarmup_fixedpreprocessing.yaml 324B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_ohlsesp_nowarmup_200epochs.yaml 322B wresnet28x10_cifar10_b128_maxlr.1_randaugn=1m=30_fixedsesp_nowarmup_8timesbatch_200epochs.yaml 322B wresnet28x10_actuallypretransformcifar10_b128_maxlr.1_ta_fixedsesp_nowarmup_200epochs_2.yaml 318B wresnet28x10_cifar100_b128_maxlr.1_uniaug_uasesp_200epochs.yaml 317B wresnet28x10_svhncore_b128_maxlr.1_ra_fixed.yaml 316B wresnet28x10_actuallypretransformcifar10_b128_maxlr.1_ta_rasesp_nowarmup_200epochs_2.yaml 312B wresnet40x2_cifar10_b128_maxlr.1_ta_subsample10_fixedsesp_nowarmup_200epochs.yaml 306B wresnet40x2_cifar10_b128_maxlr.1_ta_subsample13_fixedsesp_nowarmup_200epochs.yaml 306B wresnet40x2_cifar10_b128_maxlr.1_ta_subsample11_fixedsesp_nowarmup_200epochs.yaml 306B wresnet40x2_cifar10_b128_maxlr.1_ta_subsample12_fixedsesp_nowarmup_200epochs.yaml 306B wresnet40x2_cifar10_b128_maxlr.1_ta_subsample8_fixedsesp_nowarmup_200epochs.yaml 305B wresnet40x2_cifar10_b128_maxlr.1_ta_subsample4_fixedsesp_nowarmup_200epochs.yaml 305B wresnet28x10_svhncore_b128_maxlr.1_ta_fixedsearchspace_2_nowarmup.yaml 305B wresnet28x10_svhncore_b128_maxlr.1_ta_fixedsearchspace_2_nowarmup_fixedpreprocessing.yaml 305B wresnet40x2_cifar100_b128_maxlr.1_ta_fixedsesp_nowarmup_200epochs.yaml 303B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_fixedsesp_nowarmup_200epochs.yaml 303B wresnet40x2_cifar10_b128_maxlr.1_ta_fixedsesp_nowarmup_200epochs.yaml 302B wresnet28x10_cifar100_b128_maxlr.1_ta_fix_nowarmup_200epochs.yaml 302B wresnet28x10_svhncore_b128_maxlr.1_ta_autoaugsearchspace_2_nowarmup_fixedpreprocessing.yaml 301B shake26_2x96d_cifar10_b128_nowarmup_ta.yaml 301B wresnet28x10_actuallypretransformcifar10_b128_maxlr.1_autoaug_fixed_nowarmup_200epochs_2.yaml 300B wresnet28x10_actuallypretransformcifar10_b128_maxlr.1_autoaug_nowarmup_200epochs_2.yaml 300B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_autoaugsesp_nowarmup_200epochs.yaml 299B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_rasesp_nowarmup_200epochs.yaml 297B wresnet28x10_svhncore_b128_maxlr.1_ta_uasearchspace_2_nowarmup_fixedpreprocessing.yaml 297B wresnet28x10_cifar100_b128_maxlr.1_weighteduniaug_w=0,1_uasesp_nowarmup_200epochs.yaml 296B wresnet28x10_cifar10_b128_maxlr.1_weighteduniaug_w=0,1_uasesp_nowarmup_200epochs.yaml 295B wresnet28x10_svhncore_b128_maxlr.1_autoaug_fixedsearchspace_2_nowarmup_fixedpreprocessing.yaml 288B wresnet28x10_cifar100_b128_maxlr.1_randaug_fixedsesp_200epochs_2_nowarmup.yaml 284B wresnet28x10_cifar100_b128_maxlr.1_randaugn=1,m=30_fixedsesp_nowarmup_200epochs.yaml 284B wresnet28x10_cifar100_b128_maxlr.1_uniaug_fixedsesp_nowarmup_200epochs.yaml 283B wresnet28x10_cifar10_b128_maxlr.1_randaugn=1,m=30_fixedsesp_nowarmup_200epochs.yaml 283B wresnet40x2_cifar100_b128_maxlr.1_ua_fixedsesp_nowarmup_200epochs.yaml 282B wresnet28x10_cifar10_b128_maxlr.1_randaugn=3,m=5_fixedsesp_nowarmup_200epochs.yaml 282B wresnet28x10_cifar10_b128_maxlr.1_uniaug_fixedsesp_nowarmup_200epochs.yaml 282B wresnet40x2_cifar10_b128_maxlr.1_ua_fixedsesp_nowarmup_200epochs.yaml 281B wresnet28x10_svhncore_b128_maxlr.1_ra_n=1m=30_fixed_nowarmup.yaml 275B wresnet28x10_cifar100_b128_maxlr.1_uniaug_uasesp_nowarmup_200epochs.yaml 275B wresnet28x10_svhncore_b128_maxlr.1_ra_fixed_nowarmup.yaml 274B wresnet40x2_cifar100_b128_maxlr.1_ua_uasesp_nowarmup_200epochs.yaml 274B wresnet40x2_cifar10_b128_maxlr.1_ua_uasesp_nowarmup_200epochs.yaml 273B wresnet28x10_cifar10_b128_maxlr.1.yaml 272B wresnet28x10_cifar100_b128_maxlr.1_autoaug_nowarmup_200epochs.yaml 244B wresnet28x10_cifar100_b128_maxlr.1_faa_nowarmup_200epochs.yaml 244B wresnet28x10_cifar10_b128_maxlr.1_autoaug_fixed_nowarmup_200epochs.yaml 243B wresnet28x10_cifar10_b128_maxlr.1_autoaug_nowarmup_200epochs.yaml 243B wresnet28x10_cifar10_b128_maxlr.1_faa_nowarmup_200epochs.yaml 243B wresnet40x2_cifar100_b128_maxlr.1_faa_nowarmup_200epochs.yaml 243B wresnet40x2_cifar10_b128_maxlr.1_faa_nowarmup_200epochs.yaml 242B shake26_2x96d_cifar10_b128_nowarmup.yaml 238B共 100 条
- 1
资源评论
