IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[POT] Precommit reference update (#12304)

Browse Source 
* WIP graph tests fixing

* Fix collectiors graph tests

* remove debug code

* Fix rebase

* eps update for scales tests

* Outputs for some reference models was changed

* Sanity reference metrics update for VNNI CI hosts

* Unused hyperopt dependency which broke python3.6 support is commented

* Minor comments fixes
This commit is contained in:
Daniil Lyakhov 2022-08-05 15:50:45 +02:00 committed by GitHub
parent 3d337ccbbb
commit edaa7f8e7e
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
43 changed files with 145 additions and 139 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
tools/pot/setup.py
View File

@ -78,7 +78,7 @@ INSTALL_REQUIRES = [
] ]
ALGO_EXTRAS = [ ALGO_EXTRAS = [
'hyperopt~=0.1.2', #'hyperopt~=0.1.2',
] ]
DEV_EXTRAS = ['pytest==4.5.0', 'openpyxl==2.6.4', 'pytest-mock==3.1.1'] DEV_EXTRAS = ['pytest==4.5.0', 'openpyxl==2.6.4', 'pytest-mock==3.1.1']

4
tools/pot/tests/data/reference_fake_quantize_conf/mobilenet-v2-pytorch_asymmetric.json
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:9b8ba1d075ecba49ff3b58f98d2b05cb72a2093099dcfe0c9c884ec603595799 oid sha256:4eacc375e5cf078c565e51240a081ba435d49969a58fc5b8342738063aaaaa19
size 899722 size 900100

4
tools/pot/tests/data/reference_fake_quantize_conf/mobilenet-v2-pytorch_mixed.json
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:69eca9d6dfb46c8da845d111a0a81a164ae509aa7a39c2669c8ca0c0d9875c06 oid sha256:a8c1fafaaf950a72fcd8beb58ea8e800e27f32531456c866a5589ea54aa156d1
size 946760 size 947132

4
tools/pot/tests/data/reference_fake_quantize_conf/mobilenet-v2-pytorch_symmetric.json
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:af1e1cb69f62015c936c5a0295bdd84cc287496325901a3f4eb920ef3def5897 oid sha256:fd3a61acfdcbd30b1aab0311ba022abaed2cdbbe5d7537de3f14871a6adf1af5
size 942973 size 943354

4
tools/pot/tests/data/reference_fake_quantize_conf/resnet-50-pytorch_asymmetric.json
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:ba1ab71c22d6f7e4a57320c4dd2f4d455f5c6b58dd80ff0a04b0c5bd874da380 oid sha256:7e8453277913ce5e6248474c6939ad1ea85e4a7fb9fe640291585818b07167d0
size 1172677 size 1173302

4
tools/pot/tests/data/reference_fake_quantize_conf/resnet-50-pytorch_mixed.json
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:39109ed61b27f1ea9f993e05eefd0302fc2fa24f8a5d05164ae1cec9f3308554 oid sha256:8cf328ee30998be2ff28da6c09399a161f5d997c4ef19e574afaad0e4cf179fd
size 1167073 size 1167698

4
tools/pot/tests/data/reference_fake_quantize_conf/resnet-50-pytorch_symmetric.json
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:b832c6e49ed036981878ebfe1a4d4ed5561efdb113d927c93bb5e2fbb8e911d1 oid sha256:b195b543ae735758a21f986e91599be5caff4bba023bc5515560e6b0467f78f7
size 1167072 size 1167697

4
tools/pot/tests/data/reference_models/densenet121_example_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:4daf9a0f4cc705d8c57d7d75ad10f614d8948e1c19bb32a7b3ec2c3925f61bbb oid sha256:9647758c60dcb7a52dd52d18180c16c4d9d730c52312485cf7a53c469344e93b
size 1114481 size 1124823

4
tools/pot/tests/data/reference_models/googlenet_example_cut_fq_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:3e344bd08e82f6724c35b18a1ecac544cac637507fe40a179fc1d5f897aa4d6e oid sha256:009d181fbd73b29e09b54913779614648aa2bf896c855a5eba7a3d622047b139
size 255152 size 257031

4
tools/pot/tests/data/reference_models/googlenet_example_ig_params_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:271c108b8f91a7c396ba81deba21c4453a5a1491baa5db645b26222c801f9193 oid sha256:3452aaf55d5c0fc5e385ce1039539114f8f92cc804110497e1b83b4e03871c52
size 281049 size 282989

4
tools/pot/tests/data/reference_models/googlenet_example_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:0e2e65663a869c63b6f52b8eb6eacb3af95f8a7280881ec5626e9e882d2612ae oid sha256:ef3bbc711f59adf0ba3a2988f06ade0bc7f1483f27bfdd313778b51c244ac12c
size 282650 size 284646

4
tools/pot/tests/data/reference_models/mobilenet_v3_small_example_stat_mean_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:2194d074fe8bd0f9af9bc9ac032a81aa25f430212e81a1ae58047523a4638ea1 oid sha256:21d9880945b2f0b8df18d6815d93f7d6fc3a0facb48e5fcea2ab788197a74d95
size 223310 size 226290

4
tools/pot/tests/data/reference_models/mobilenetv2_example_cut_fq_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:97178199c399898b94db5dece4be4f5a80df1a33a6c46d21092da7eb5c56c7dc oid sha256:40b3cf5ff3cb1bcb458333c91c50c8d1ad3d9f564450af7d7a66cfad601ea8e2
size 260217 size 261369

4
tools/pot/tests/data/reference_models/mobilenetv2_example_ig_params_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:9068d0230ba67e41fbbe81d2e6a65b2ded829ada133ea13b440298d1fa653cf7 oid sha256:5be503e58809d19cd32304547084d4564e289e43f297067bafebbe013e04742d
size 256177 size 257372

4
tools/pot/tests/data/reference_models/mobilenetv2_example_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:10af97d1859baacd6072c6362a0de2cf506e57e739feeecca4eb710d18da86e3 oid sha256:b59650097bce8071c9f3b107b30f0700616e88bf23e0657c84d33c2ad056bb7a
size 272339 size 273580

4
tools/pot/tests/data/reference_models/mobilenetv2_example_stat_mean_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:7bc10311127ca297307f145663e6b2b516ae1c6e187603e6a3941df75a8c90cd oid sha256:90e2fa5a7479eee8798b6dfb165d30cd4fd482c1096862774cbe7e21eba22831
size 112888 size 113812

4
tools/pot/tests/data/reference_models/mobilenetv2_ssd_example_cut_fq_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:c023025650f024b5f8ac4d08f24a085c822eaeeaf34de5c5fdf5683d12d89537 oid sha256:f11cead05263a2852516e10ee89ec6b2d1063ffe35c18af265dc0905602e850d
size 423780 size 426443

4
tools/pot/tests/data/reference_models/mobilenetv2_ssd_example_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:cce827274e90ae46c088c3e7a7825657e7bd9fe8451f61d3c587400a820e9fe1 oid sha256:ea3a90ccc70d4d741b1d067be92ae192389fa39bac0452a0551ec67fc267d569
size 435555 size 438269

4
tools/pot/tests/data/reference_models/mobilenetv2_ssd_example_stat_mean_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:50fbab2ede57a50135d5367ad0b5a85409386a2273564ce831eb7572a33e6f29 oid sha256:c81ae2a2558951f131064301028daf3b0fc652ce9fbbd63d22079b336cbdf108
size 325967 size 330017

4
tools/pot/tests/data/reference_models/mtcnn_ig_params_onet_caffe.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:4cbb5506a2aa95c3dfdba0f2e6f3c7a69eb4879504d499077805718f280834f1 oid sha256:f476ebfa8f7dc51422a04f091db0bd0df6cb25624811825846b9fd07969ca5f2
size 48464 size 48457

4
tools/pot/tests/data/reference_models/mtcnn_ig_params_pnet_caffe.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:4087cdc57c73e461d6341f1493f1c6dbe87350cab8477942aea3b4fba591f735 oid sha256:485fd47db7285675e6ae418e2125c6d45646be0571e4fffc6f3ae15b71d72999
size 30334 size 30329

4
tools/pot/tests/data/reference_models/mtcnn_ig_params_rnet_caffe.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:66cd71d8f24f840b8840223a811deab944ea54e25398befba9498ac0daed1da1 oid sha256:764e4ed2b2394ba76e38b15d6e8c7c4e12057bffc133b332ea67be3e75739957
size 28059 size 28035

4
tools/pot/tests/data/reference_models/multiple_out_ports_net_cut_fq_tf.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:7928108e04c45c786210abfabca768a8ae7757a0338537e7ebafac11f2fe3c11 oid sha256:de7dc61a5a81f0b8d0f5afc272adc62157cc4b0b885506c8d176e1c4905b18f5
size 10588 size 10558

4
tools/pot/tests/data/reference_models/multiple_out_ports_net_tf.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:bdcd4a2a09316358d1adb04ad1d5c0ff14ad065ffe0bf9fd437c2853c3f83cb9 oid sha256:45303a7659cdfd9f434c317e23792b5456dd2c4b1186b5fb207e9b4056dfd64f
size 9660 size 10558

4
tools/pot/tests/data/reference_models/resnet_example_ig_params_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:7dabca07167805c3d7286b522fb5fbc880e1a652b7fa41ccd01243aa6aa6003f oid sha256:aeb687993302424a51747a12c79fcd241b49d1c6b69c2775550b4b42bf6f4f14
size 103074 size 103816

4
tools/pot/tests/data/reference_models/resnet_example_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:9015cfdbdd0acf569b83ff38896750b9aba2d4056ad942458e954972b454c1ec oid sha256:ef7c4f5fdfc04ec0b5d0091310682cc9ff1b9f3ebabdfb9f5c33f056bb7adcec
size 120908 size 121728

4
tools/pot/tests/data/reference_models/resnet_example_stat_abs_max_max_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:75b7b8462348bc60384493d9f8094a7ac17240b3aa10c651378308fab0fd1f42 oid sha256:53d6f54ffc9c77a003c3f6bc38acedcba19068b2cfa6fedcef72c522ad2ff53a
size 109305 size 110895

4
tools/pot/tests/data/reference_models/resnet_example_stat_abs_max_min_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:f6bf2c492d93d7a2c23b2c3ebd7a76524d04504fffecd41ae694a85844bc226c oid sha256:0b2ad81c25ccd993aba8ad724357a0bbf3c22fabeedd315a66de92551559e816
size 111563 size 113153

4
tools/pot/tests/data/reference_models/resnet_example_stat_max_min_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:c35dcb3585d82a74d9fd5cbe82940e66b2710786979e51a27bce53e3d66a82fd oid sha256:c38e890cb0842e1f83366ee6a59626450499cfc01c8f3072be20253fa26f2490
size 101358 size 102836

4
tools/pot/tests/data/reference_models/resnet_example_stat_min_quantile_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:dfeffb1a15e39d5942ef61d9ece68458a961209831556fa1fe0a3a08ca32c5f1 oid sha256:2eeffa3d107dc322b46d7b5d78988357248966e287a70ebd47aaaa585ddb57f5
size 76579 size 77612

4
tools/pot/tests/data/reference_models/resnet_example_stat_quantile_abs_quantile_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:8024be4a3e32e9567b9fccb4b2001b90e1d2e7cd2a634ea0565cbf845ac476a0 oid sha256:eca87ed59e6bece6f9d6e2d38e80e6ac8892b2e34df3edeeee0f1c030ae239ed
size 54066 size 54654

4
tools/pot/tests/data/reference_models/squeezenet1_1_example_cut_fq_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:670840ab03dbe297d0cc1a7aa8f2586612aa39559df1989ef99ac91fa61c0fa8 oid sha256:1cb38a61f982c19e411b62654da4a72c8163730d9d09d72b6ca34de623be2fdd
size 149416 size 150577

4
tools/pot/tests/data/reference_models/squeezenet1_1_example_stat_mean_pytorch.xml
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:f96a3afe2b788ae4e460c57a4ed25da10883a553439e27ce22a726a5d4075f20 oid sha256:156e32d7a9e650a65bb4e7a4debf9378c80ec760694dae87e36a1a6ee3ce315c
size 125489 size 127384

2
tools/pot/tests/data/reference_scale/mobilenet-v2-pytorch_accuracy_activations.json
View File

@ -1 +1 @@
{"534/fq_input_0": [0.0, 0.049394913017749786], "478/fq_input_0": [-0.015164043754339218, 0.017059549689292908], "478/fq_input_1": [-0.018027909100055695, 0.014575756154954433], "Convolution_1461/fq_input_0": [-0.03508269414305687, 0.04339175671339035], "GroupConvolution_458/fq_input_0": [0.0, 0.015301208011806011], "465/fq_input_0": [-0.013670912012457848, 0.015872159972786903], "465/fq_input_1": [-0.009457818232476711, 0.012467123568058014], "351/fq_input_0": [-0.026998696848750114, 0.02678776904940605], "351/fq_input_1": [-0.041718490421772, 0.04204697906970978], "GroupConvolution_3843/fq_input_0": [-1.0507744718779577e-06, 0.0006702778628095984], "GroupConvolution_1811/fq_input_0": [0.0, 0.002688088221475482], "Convolution_3895/fq_input_0": [0.0, 0.010443663224577904], "Convolution_285/fq_input_0": [0.0, 0.049394913017749786], "GroupConvolution_1134/fq_input_0": [0.0, 0.004058439284563065], "503/fq_input_0": [-0.014566807076334953, 0.015149478800594807], "503/fq_input_1": [-0.007149680983275175, 0.00830090045928955], "Convolution_333/fq_input_0": [-0.0555744506418705, 0.04786704108119011], "Convolution_1863/fq_input_0": [0.0, 0.011041364632546902], "516/fq_input_0": [-0.013903681188821793, 0.017471883445978165], "516/fq_input_1": [-0.022818375378847122, 0.02228771522641182], "Convolution_2992/fq_input_0": [0.0, 0.017005037516355515], "Convolution_3041/fq_input_0": [-0.022879013791680336, 0.02698550373315811], "Convolution_1412/fq_input_0": [0.0, 0.014988739974796772], "440/fq_input_0": [-0.025955941528081894, 0.02098565548658371], "440/fq_input_1": [-0.02956792712211609, 0.01612796075642109], "Convolution_108/fq_input_0": [-0.9615384936332703, 1.0], "GroupConvolution_3166/fq_input_0": [-2.304807722453006e-08, 0.004747476428747177], "Convolution_2089/fq_input_0": [0.0, 0.011155050247907639], "427/fq_input_0": [-0.02210625447332859, 0.021256014704704285], "427/fq_input_1": [-0.012997206300497055, 0.011372555047273636], "GroupConvolution_3617/fq_input_0": [-2.765769302470744e-08, 0.0025624139234423637], "Convolution_3218/fq_input_0": [0.0, 0.020942626520991325], "Convolution_735/fq_input_0": [0.0, 0.022948624566197395], "GroupConvolution_909/fq_input_0": [0.0, 0.009065578691661358], "GroupConvolution_2940/fq_input_0": [-8.066827916763941e-08, 0.0026795677840709686], "GroupConvolution_2489/fq_input_0": [0.0, 0.00370213994756341], "376/fq_input_0": [-0.025736434385180473, 0.022876830771565437], "376/fq_input_1": [-0.02580964006483555, 0.02294190227985382], "389/fq_input_0": [-0.03178860992193222, 0.03690711781382561], "389/fq_input_1": [-0.01618143543601036, 0.01764041744172573], "414/fq_input_0": [-0.02402852103114128, 0.01822853460907936], "414/fq_input_1": [-0.014876025728881359, 0.018993496894836426], "GroupConvolution_233/fq_input_0": [-8.297308227156464e-07, 0.007950618863105774], "Convolution_2364/fq_input_0": [-0.030782364308834076, 0.024887869134545326], "GroupConvolution_683/fq_input_0": [0.0, 0.005861522629857063], "GroupConvolution_3391/fq_input_0": [-1.382884651235372e-08, 0.0023272547405213118], "Convolution_2766/fq_input_0": [0.0, 0.02053709141910076], "Convolution_2541/fq_input_0": [0.0, 0.018151046708226204], "Convolution_2315/fq_input_0": [0.0, 0.027796654030680656], "GroupConvolution_1360/fq_input_0": [-3.457211406043825e-08, 0.004260015673935413], "GroupConvolution_2037/fq_input_0": [-5.18581764197279e-08, 0.0026650403160601854], "Convolution_3669/fq_input_0": [0.0, 0.01576113887131214], "GroupConvolution_1586/fq_input_0": [0.0, 0.006106846034526825], "Convolution_1638/fq_input_0": [0.0, 0.014108488336205482], "Convolution_961/fq_input_0": [0.0, 0.020471246913075447], "Convolution_3943/fq_input_0": [-0.012309478595852852, 0.009795570746064186], "GroupConvolution_2263/fq_input_0": [-1.037163528394558e-07, 0.0025707946624606848], "Convolution_784/fq_input_0": [-0.04988809674978256, 0.05188361927866936], "Convolution_3443/fq_input_0": [0.0, 0.020869001746177673], "Convolution_1186/fq_input_0": [0.0, 0.014912507496774197], "GroupConvolution_2714/fq_input_0": [-8.066827206221205e-08, 0.002713250694796443], "Convolution_510/fq_input_0": [0.0, 0.02854330465197563], "Convolution_3718/fq_input_0": [-0.028473246842622757, 0.035216908901929855], "542/fq_input_0": [0.0, 0.03368273004889488]} {"Conv_148/WithoutBiases/fq_input_0": [-2.765769302470744e-08, 0.0025624153204262257], "Add_56/fq_input_0": [-0.03178860619664192, 0.03690711408853531], "Add_56/fq_input_1": [-0.01618143357336521, 0.01764041557908058], "Conv_125/WithoutBiases/fq_input_0": [-0.022879047319293022, 0.026985542848706245], "Conv_84/WithoutBiases/fq_input_0": [0.0, 0.011155051179230213], "Conv_162/WithoutBiases/fq_input_0": [0.0, 0.010443665087223053], "Conv_61/WithoutBiases/fq_input_0": [0.0, 0.006106846500188112], "Conv_119/WithoutBiases/fq_input_0": [-8.066827916763941e-08, 0.002679567551240325], "Conv_57/WithoutBiases/fq_input_0": [-0.03508269414305687, 0.04339175671339035], "Conv_154/WithoutBiases/fq_input_0": [-0.028473293408751488, 0.03521696850657463], "Conv_55/WithoutBiases/fq_input_0": [0.0, 0.014988746494054794], "Conv_0/WithoutBiases/fq_input_0": [-0.9615384936332703, 1.0], "Conv_163/WithoutBiases/fq_input_0": [-0.01230948232114315, 0.00979557354003191], "Conv_32/WithoutBiases/fq_input_0": [0.0, 0.009065578691661358], "Conv_45/WithoutBiases/fq_input_0": [0.0, 0.014912506565451622], "Add_114/fq_input_0": [-0.013670899905264378, 0.01587214693427086], "Add_114/fq_input_1": [-0.009457815438508987, 0.01246712077409029], "Conv_113/WithoutBiases/fq_input_0": [0.0, 0.020537080243229866], "Conv_90/WithoutBiases/fq_input_0": [-1.037163528394558e-07, 0.0025707946624606848], "Conv_8/WithoutBiases/fq_input_0": [0.0, 0.049394913017749786], "Conv_138/WithoutBiases/fq_input_0": [-1.382884651235372e-08, 0.0023272561375051737], "Conv_104/WithoutBiases/fq_input_0": [0.0, 0.018151046708226204], "Conv_36/WithoutBiases/fq_input_0": [0.0, 0.020471245050430298], "Conv_51/WithoutBiases/fq_input_0": [-3.457211406043825e-08, 0.004260015673935413], "Conv_28/WithoutBiases/fq_input_0": [-0.04988810047507286, 0.051883623003959656], "Add_153/fq_input_0": [-0.013903711922466755, 0.01747192069888115], "Add_153/fq_input_1": [-0.022818440571427345, 0.022287780418992043], "Conv_70/WithoutBiases/fq_input_0": [0.0, 0.002688088221475482], "Add_95/fq_input_0": [-0.025955934077501297, 0.020985649898648262], "Add_95/fq_input_1": [-0.02956792712211609, 0.01612795889377594], "Add_124/fq_input_0": [-0.015164051204919815, 0.017059559002518654], "Add_124/fq_input_1": [-0.01802794076502323, 0.014575780369341373], "Conv_26/WithoutBiases/fq_input_0": [0.0, 0.022948624566197395], "Conv_80/WithoutBiases/fq_input_0": [-5.18581764197279e-08, 0.002665039850398898], "Conv_109/WithoutBiases/fq_input_0": [-8.066827206221205e-08, 0.0027132511604577303], "Conv_17/WithoutBiases/fq_input_0": [0.0, 0.028543302789330482], "Conv_152/WithoutBiases/fq_input_0": [0.0, 0.015761131420731544], "GlobalAveragePool_167/fq_input_0": [0.0, 0.049394913017749786], "Conv_9/WithoutBiases/fq_input_0": [-0.0555744469165802, 0.04786704108119011], "Conv_65/WithoutBiases/fq_input_0": [0.0, 0.014108490198850632], "Conv_133/WithoutBiases/fq_input_0": [0.0, 0.02094263955950737], "Conv_13/WithoutBiases/fq_input_0": [0.0, 0.015301208011806011], "Conv_94/WithoutBiases/fq_input_0": [0.0, 0.027796650305390358], "Add_75/fq_input_0": [-0.024028511717915535, 0.018228525295853615], "Add_75/fq_input_1": [-0.014876028522849083, 0.018993500620126724], "Conv_74/WithoutBiases/fq_input_0": [0.0, 0.011041362769901752], "Conv_142/WithoutBiases/fq_input_0": [0.0, 0.02086900733411312], "Conv_22/WithoutBiases/fq_input_0": [0.0, 0.005861522629857063], "Conv_129/WithoutBiases/fq_input_0": [-2.304807722453006e-08, 0.004747480154037476], "Conv_123/WithoutBiases/fq_input_0": [0.0, 0.017005059868097305], "Add_85/fq_input_0": [-0.02210625633597374, 0.021256016567349434], "Add_85/fq_input_1": [-0.012997213751077652, 0.011372562497854233], "Add_143/fq_input_0": [-0.014566809870302677, 0.01514948159456253], "Add_143/fq_input_1": [-0.007149683777242899, 0.00830090418457985], "Add_46/fq_input_0": [-0.025736436247825623, 0.022876832634210587], "Add_46/fq_input_1": [-0.025809651240706444, 0.022941911593079567], "Conv_96/WithoutBiases/fq_input_0": [-0.030782364308834076, 0.024887869134545326], "Conv_41/WithoutBiases/fq_input_0": [0.0, 0.004058439284563065], "Conv_100/WithoutBiases/fq_input_0": [0.0, 0.00370213994756341], "Conv_158/WithoutBiases/fq_input_0": [-1.0507754950594972e-06, 0.0006702779210172594], "Add_27/fq_input_0": [-0.026998700574040413, 0.02678777277469635], "Add_27/fq_input_1": [-0.041718482971191406, 0.04204697161912918], "Conv_4/WithoutBiases/fq_input_0": [-8.297308227156464e-07, 0.007950618863105774], "Reshape_173/fq_input_0": [0.0, 0.033682722598314285]}

2
tools/pot/tests/data/reference_scale/mobilenet-v2-pytorch_clipped_activations.json
View File

@ -1 +1 @@
{"534/fq_input_0": [0.0, 0.024504506960511208], "478/fq_input_0": [-0.016956167295575142, 0.016823697835206985], "478/fq_input_1": [-0.017891479656100273, 0.017751703038811684], "Convolution_1461/fq_input_0": [-0.04339175671339035, 0.024504506960511208], "GroupConvolution_458/fq_input_0": [0.0, 0.01409971434623003], "465/fq_input_0": [-0.015748171135783195, 0.015625139698386192], "465/fq_input_1": [-0.012467123568058014, 0.012369723990559578], "351/fq_input_0": [-0.026998037472367287, 0.024504506960511208], "351/fq_input_1": [-0.04204697906970978, 0.024504506960511208], "GroupConvolution_3843/fq_input_0": [0.0, 0.0006734440103173256], "GroupConvolution_1811/fq_input_0": [0.0, 0.002667087595909834], "Convolution_3895/fq_input_0": [0.0, 0.010362071916460991], "Convolution_285/fq_input_0": [0.0, 0.024504506960511208], "GroupConvolution_1134/fq_input_0": [0.0, 0.004026732407510281], "503/fq_input_0": [-0.01505535189062357, 0.014937731437385082], "503/fq_input_1": [-0.008267928846180439, 0.008203336037695408], "Convolution_333/fq_input_0": [-0.0555744469165802, 0.024504506960511208], "Convolution_1863/fq_input_0": [0.0, 0.010955104604363441], "516/fq_input_0": [-0.017471881583333015, 0.017335383221507072], "516/fq_input_1": [-0.022786367684602737, 0.022608349099755287], "Convolution_2992/fq_input_0": [0.0, 0.016872184351086617], "Convolution_3041/fq_input_0": [-0.026985501870512962, 0.024504506960511208], "Convolution_1412/fq_input_0": [0.0, 0.014871640130877495], "440/fq_input_0": [-0.025897065177559853, 0.024504506960511208], "440/fq_input_1": [-0.0293196402490139, 0.024504506960511208], "Convolution_108/fq_input_0": [-1.0, 0.024504506960511208], "GroupConvolution_3166/fq_input_0": [0.0, 0.004710768815129995], "Convolution_2089/fq_input_0": [0.0, 0.01106790080666542], "427/fq_input_0": [-0.022085215896368027, 0.021912673488259315], "427/fq_input_1": [-0.01299720536917448, 0.012895665131509304], "GroupConvolution_3617/fq_input_0": [0.0, 0.0025426887441426516], "Convolution_3218/fq_input_0": [0.0, 0.020779011771082878], "Convolution_735/fq_input_0": [0.0, 0.022769339382648468], "GroupConvolution_909/fq_input_0": [0.0, 0.008994753472507], "GroupConvolution_2940/fq_input_0": [0.0, 0.002659756224602461], "GroupConvolution_2489/fq_input_0": [0.0, 0.003673217026516795], "376/fq_input_0": [-0.025736434385180473, 0.024504506960511208], "376/fq_input_1": [-0.02580964006483555, 0.024504506960511208], "389/fq_input_0": [-0.03690711781382561, 0.024504506960511208], "389/fq_input_1": [-0.017548587173223495, 0.0174114890396595], "414/fq_input_0": [-0.02402852289378643, 0.023840798065066338], "414/fq_input_1": [-0.018902117386460304, 0.018754445016384125], "GroupConvolution_233/fq_input_0": [0.0, 0.007917627692222595], "Convolution_2364/fq_input_0": [-0.03072093054652214, 0.024504506960511208], "GroupConvolution_683/fq_input_0": [0.0, 0.00581572949886322], "GroupConvolution_3391/fq_input_0": [0.0, 0.002309233183041215], "Convolution_2766/fq_input_0": [0.0, 0.020376645028591156], "Convolution_2541/fq_input_0": [0.0, 0.018009241670370102], "Convolution_2315/fq_input_0": [0.0, 0.024504506960511208], "GroupConvolution_1360/fq_input_0": [0.0, 0.004227211233228445], "GroupConvolution_2037/fq_input_0": [0.0, 0.0026447095442563295], "Convolution_3669/fq_input_0": [0.0, 0.01563800498843193], "GroupConvolution_1586/fq_input_0": [0.0, 0.006059136241674423], "Convolution_1638/fq_input_0": [0.0, 0.013998265378177166], "Convolution_961/fq_input_0": [0.0, 0.02031131461262703], "Convolution_3943/fq_input_0": [-0.012309477664530277, 0.012213310226798058], "GroupConvolution_2263/fq_input_0": [0.0, 0.00255152047611773], "Convolution_784/fq_input_0": [-0.05170396342873573, 0.024504506960511208], "Convolution_3443/fq_input_0": [0.0, 0.02070596255362034], "Convolution_1186/fq_input_0": [0.0, 0.014796003699302673], "GroupConvolution_2714/fq_input_0": [0.0, 0.0026932351756840944], "Convolution_510/fq_input_0": [0.0, 0.024504506960511208], "Convolution_3718/fq_input_0": [-0.03502354398369789, 0.024504506960511208], "542/fq_input_0": [0.0, 0.024504506960511208]} {"Conv_148/WithoutBiases/fq_input_0": [0.0, 0.002542690373957157], "Add_56/fq_input_0": [-0.03690711408853531, 0.024504506960511208], "Add_56/fq_input_1": [-0.017548592761158943, 0.017411494627594948], "Conv_125/WithoutBiases/fq_input_0": [-0.026985540986061096, 0.024504506960511208], "Conv_84/WithoutBiases/fq_input_0": [0.0, 0.011067901737987995], "Conv_162/WithoutBiases/fq_input_0": [0.0, 0.01036207377910614], "Conv_61/WithoutBiases/fq_input_0": [0.0, 0.0060591367073357105], "Conv_119/WithoutBiases/fq_input_0": [0.0, 0.002659756224602461], "Conv_57/WithoutBiases/fq_input_0": [-0.04339175671339035, 0.024504506960511208], "Conv_154/WithoutBiases/fq_input_0": [-0.03502371534705162, 0.024504506960511208], "Conv_55/WithoutBiases/fq_input_0": [0.0, 0.014871646650135517], "Conv_0/WithoutBiases/fq_input_0": [-1.0, 0.024504506960511208], "Conv_163/WithoutBiases/fq_input_0": [-0.012309481389820576, 0.012213313952088356], "Conv_32/WithoutBiases/fq_input_0": [0.0, 0.008994753472507], "Conv_45/WithoutBiases/fq_input_0": [0.0, 0.014796002767980099], "Add_114/fq_input_0": [-0.015748165547847748, 0.015625134110450745], "Add_114/fq_input_1": [-0.01246712077409029, 0.012369721196591854], "Conv_113/WithoutBiases/fq_input_0": [0.0, 0.02037663385272026], "Conv_90/WithoutBiases/fq_input_0": [0.0, 0.002551520708948374], "Conv_8/WithoutBiases/fq_input_0": [0.0, 0.024504506960511208], "Conv_138/WithoutBiases/fq_input_0": [0.0, 0.002309234580025077], "Conv_104/WithoutBiases/fq_input_0": [0.0, 0.018009241670370102], "Conv_36/WithoutBiases/fq_input_0": [0.0, 0.02031131274998188], "Conv_51/WithoutBiases/fq_input_0": [0.0, 0.004227211698889732], "Conv_28/WithoutBiases/fq_input_0": [-0.05170397087931633, 0.024504506960511208], "Add_153/fq_input_0": [-0.01747192069888115, 0.017335420474410057], "Add_153/fq_input_1": [-0.02278638817369938, 0.022608371451497078], "Conv_70/WithoutBiases/fq_input_0": [0.0, 0.002667087595909834], "Add_95/fq_input_0": [-0.0258970744907856, 0.024504506960511208], "Add_95/fq_input_1": [-0.0293196439743042, 0.024504506960511208], "Add_124/fq_input_0": [-0.016956161707639694, 0.01682369038462639], "Add_124/fq_input_1": [-0.01789148710668087, 0.01775171048939228], "Conv_26/WithoutBiases/fq_input_0": [0.0, 0.022769339382648468], "Conv_80/WithoutBiases/fq_input_0": [0.0, 0.0026447095442563295], "Conv_109/WithoutBiases/fq_input_0": [0.0, 0.0026932351756840944], "Conv_17/WithoutBiases/fq_input_0": [0.0, 0.024504506960511208], "Conv_152/WithoutBiases/fq_input_0": [0.0, 0.015637997537851334], "GlobalAveragePool_167/fq_input_0": [0.0, 0.024504506960511208], "Conv_9/WithoutBiases/fq_input_0": [-0.0555744431912899, 0.024504506960511208], "Conv_65/WithoutBiases/fq_input_0": [0.0, 0.013998267240822315], "Conv_133/WithoutBiases/fq_input_0": [0.0, 0.020779024809598923], "Conv_13/WithoutBiases/fq_input_0": [0.0, 0.01409971434623003], "Conv_94/WithoutBiases/fq_input_0": [0.0, 0.024504506960511208], "Add_75/fq_input_0": [-0.024028509855270386, 0.02384078875184059], "Add_75/fq_input_1": [-0.01890212669968605, 0.01875445432960987], "Conv_74/WithoutBiases/fq_input_0": [0.0, 0.010955101810395718], "Conv_142/WithoutBiases/fq_input_0": [0.0, 0.020705968141555786], "Conv_22/WithoutBiases/fq_input_0": [0.0, 0.00581572949886322], "Conv_129/WithoutBiases/fq_input_0": [0.0, 0.004710772540420294], "Conv_123/WithoutBiases/fq_input_0": [0.0, 0.016872208565473557], "Add_85/fq_input_0": [-0.02208520844578743, 0.02191266603767872], "Add_85/fq_input_1": [-0.012997213751077652, 0.012895672582089901], "Add_143/fq_input_0": [-0.015055370517075062, 0.01493775099515915], "Add_143/fq_input_1": [-0.008267931640148163, 0.008203338831663132], "Add_46/fq_input_0": [-0.025736436247825623, 0.024504506960511208], "Add_46/fq_input_1": [-0.025809651240706444, 0.024504506960511208], "Conv_96/WithoutBiases/fq_input_0": [-0.030720945447683334, 0.024504506960511208], "Conv_41/WithoutBiases/fq_input_0": [0.0, 0.004026732407510281], "Conv_100/WithoutBiases/fq_input_0": [0.0, 0.003673217026516795], "Conv_158/WithoutBiases/fq_input_0": [0.0, 0.0006734441849403083], "Add_27/fq_input_0": [-0.026998046785593033, 0.024504506960511208], "Add_27/fq_input_1": [-0.04204697161912918, 0.024504506960511208], "Conv_4/WithoutBiases/fq_input_0": [0.0, 0.007917627692222595], "Reshape_173/fq_input_0": [0.0, 0.024504506960511208]}

2
tools/pot/tests/data/reference_scale/mobilenet-v2-pytorch_performance_activations.json
View File

@ -1 +1 @@
{"534/fq_input_0": [0.0, 0.049009013921022415], "478/fq_input_0": [-0.016956167295575142, 0.016823697835206985], "478/fq_input_1": [-0.017891479656100273, 0.017751703038811684], "Convolution_1461/fq_input_0": [-0.04339175671339035, 0.043052759021520615], "GroupConvolution_458/fq_input_0": [0.0, 0.015181667171418667], "465/fq_input_0": [-0.015748171135783195, 0.015625139698386192], "465/fq_input_1": [-0.012467123568058014, 0.012369723990559578], "351/fq_input_0": [-0.026998037472367287, 0.02678711712360382], "351/fq_input_1": [-0.04204697906970978, 0.041718486696481705], "GroupConvolution_3843/fq_input_0": [0.0, 0.0006734440103173256], "GroupConvolution_1811/fq_input_0": [0.0, 0.002667087595909834], "Convolution_3895/fq_input_0": [0.0, 0.010362071916460991], "Convolution_285/fq_input_0": [0.0, 0.049009013921022415], "GroupConvolution_1134/fq_input_0": [0.0, 0.004026732407510281], "503/fq_input_0": [-0.01505535189062357, 0.014937731437385082], "503/fq_input_1": [-0.008267928846180439, 0.008203336037695408], "Convolution_333/fq_input_0": [-0.0555744469165802, 0.05514027178287506], "Convolution_1863/fq_input_0": [0.0, 0.010955104604363441], "516/fq_input_0": [-0.017471881583333015, 0.017335383221507072], "516/fq_input_1": [-0.022786367684602737, 0.022608349099755287], "Convolution_2992/fq_input_0": [0.0, 0.016872184351086617], "Convolution_3041/fq_input_0": [-0.026985501870512962, 0.026774678379297256], "Convolution_1412/fq_input_0": [0.0, 0.014871640130877495], "440/fq_input_0": [-0.025897065177559853, 0.025694744661450386], "440/fq_input_1": [-0.0293196402490139, 0.029090579599142075], "Convolution_108/fq_input_0": [-1.0, 0.9921875], "GroupConvolution_3166/fq_input_0": [0.0, 0.004710768815129995], "Convolution_2089/fq_input_0": [0.0, 0.01106790080666542], "427/fq_input_0": [-0.022085215896368027, 0.021912673488259315], "427/fq_input_1": [-0.01299720536917448, 0.012895665131509304], "GroupConvolution_3617/fq_input_0": [0.0, 0.0025426887441426516], "Convolution_3218/fq_input_0": [0.0, 0.020779011771082878], "Convolution_735/fq_input_0": [0.0, 0.022769339382648468], "GroupConvolution_909/fq_input_0": [0.0, 0.008994753472507], "GroupConvolution_2940/fq_input_0": [0.0, 0.002659756224602461], "GroupConvolution_2489/fq_input_0": [0.0, 0.003673217026516795], "376/fq_input_0": [-0.025736434385180473, 0.02553536742925644], "376/fq_input_1": [-0.02580964006483555, 0.0256080012768507], "389/fq_input_0": [-0.03690711781382561, 0.036618780344724655], "389/fq_input_1": [-0.017548587173223495, 0.0174114890396595], "414/fq_input_0": [-0.02402852289378643, 0.023840798065066338], "414/fq_input_1": [-0.018902117386460304, 0.018754445016384125], "GroupConvolution_233/fq_input_0": [0.0, 0.007917627692222595], "Convolution_2364/fq_input_0": [-0.03072093054652214, 0.030480923131108284], "GroupConvolution_683/fq_input_0": [0.0, 0.00581572949886322], "GroupConvolution_3391/fq_input_0": [0.0, 0.002309233183041215], "Convolution_2766/fq_input_0": [0.0, 0.020376645028591156], "Convolution_2541/fq_input_0": [0.0, 0.018009241670370102], "Convolution_2315/fq_input_0": [0.0, 0.027579491958022118], "GroupConvolution_1360/fq_input_0": [0.0, 0.004227211233228445], "GroupConvolution_2037/fq_input_0": [0.0, 0.0026447095442563295], "Convolution_3669/fq_input_0": [0.0, 0.01563800498843193], "GroupConvolution_1586/fq_input_0": [0.0, 0.006059136241674423], "Convolution_1638/fq_input_0": [0.0, 0.013998265378177166], "Convolution_961/fq_input_0": [0.0, 0.02031131461262703], "Convolution_3943/fq_input_0": [-0.012309477664530277, 0.012213310226798058], "GroupConvolution_2263/fq_input_0": [0.0, 0.00255152047611773], "Convolution_784/fq_input_0": [-0.05170396342873573, 0.05130002275109291], "Convolution_3443/fq_input_0": [0.0, 0.02070596255362034], "Convolution_1186/fq_input_0": [0.0, 0.014796003699302673], "GroupConvolution_2714/fq_input_0": [0.0, 0.0026932351756840944], "Convolution_510/fq_input_0": [0.0, 0.0283203087747097], "Convolution_3718/fq_input_0": [-0.03502354398369789, 0.03474992513656616], "542/fq_input_0": [0.0, 0.03341958299279213]} {"Conv_148/WithoutBiases/fq_input_0": [0.0, 0.002542690373957157], "Add_56/fq_input_0": [-0.03690711408853531, 0.03661877661943436], "Add_56/fq_input_1": [-0.017548592761158943, 0.017411494627594948], "Conv_125/WithoutBiases/fq_input_0": [-0.026985540986061096, 0.02677471749484539], "Conv_84/WithoutBiases/fq_input_0": [0.0, 0.011067901737987995], "Conv_162/WithoutBiases/fq_input_0": [0.0, 0.01036207377910614], "Conv_61/WithoutBiases/fq_input_0": [0.0, 0.0060591367073357105], "Conv_119/WithoutBiases/fq_input_0": [0.0, 0.002659756224602461], "Conv_57/WithoutBiases/fq_input_0": [-0.04339175671339035, 0.043052759021520615], "Conv_154/WithoutBiases/fq_input_0": [-0.03502371534705162, 0.03475009277462959], "Conv_55/WithoutBiases/fq_input_0": [0.0, 0.014871646650135517], "Conv_0/WithoutBiases/fq_input_0": [-1.0, 0.9921875], "Conv_163/WithoutBiases/fq_input_0": [-0.012309481389820576, 0.012213313952088356], "Conv_32/WithoutBiases/fq_input_0": [0.0, 0.008994753472507], "Conv_45/WithoutBiases/fq_input_0": [0.0, 0.014796002767980099], "Add_114/fq_input_0": [-0.015748165547847748, 0.015625134110450745], "Add_114/fq_input_1": [-0.01246712077409029, 0.012369721196591854], "Conv_113/WithoutBiases/fq_input_0": [0.0, 0.02037663385272026], "Conv_90/WithoutBiases/fq_input_0": [0.0, 0.002551520708948374], "Conv_8/WithoutBiases/fq_input_0": [0.0, 0.049009013921022415], "Conv_138/WithoutBiases/fq_input_0": [0.0, 0.002309234580025077], "Conv_104/WithoutBiases/fq_input_0": [0.0, 0.018009241670370102], "Conv_36/WithoutBiases/fq_input_0": [0.0, 0.02031131274998188], "Conv_51/WithoutBiases/fq_input_0": [0.0, 0.004227211698889732], "Conv_28/WithoutBiases/fq_input_0": [-0.05170397087931633, 0.051300033926963806], "Add_153/fq_input_0": [-0.01747192069888115, 0.017335420474410057], "Add_153/fq_input_1": [-0.02278638817369938, 0.022608371451497078], "Conv_70/WithoutBiases/fq_input_0": [0.0, 0.002667087595909834], "Add_95/fq_input_0": [-0.0258970744907856, 0.025694752112030983], "Add_95/fq_input_1": [-0.0293196439743042, 0.029090583324432373], "Add_124/fq_input_0": [-0.016956161707639694, 0.01682369038462639], "Add_124/fq_input_1": [-0.01789148710668087, 0.01775171048939228], "Conv_26/WithoutBiases/fq_input_0": [0.0, 0.022769339382648468], "Conv_80/WithoutBiases/fq_input_0": [0.0, 0.0026447095442563295], "Conv_109/WithoutBiases/fq_input_0": [0.0, 0.0026932351756840944], "Conv_17/WithoutBiases/fq_input_0": [0.0, 0.028320306912064552], "Conv_152/WithoutBiases/fq_input_0": [0.0, 0.015637997537851334], "GlobalAveragePool_167/fq_input_0": [0.0, 0.049009013921022415], "Conv_9/WithoutBiases/fq_input_0": [-0.0555744431912899, 0.05514026805758476], "Conv_65/WithoutBiases/fq_input_0": [0.0, 0.013998267240822315], "Conv_133/WithoutBiases/fq_input_0": [0.0, 0.020779024809598923], "Conv_13/WithoutBiases/fq_input_0": [0.0, 0.015181667171418667], "Conv_94/WithoutBiases/fq_input_0": [0.0, 0.02757948823273182], "Add_75/fq_input_0": [-0.024028509855270386, 0.02384078875184059], "Add_75/fq_input_1": [-0.01890212669968605, 0.01875445432960987], "Conv_74/WithoutBiases/fq_input_0": [0.0, 0.010955101810395718], "Conv_142/WithoutBiases/fq_input_0": [0.0, 0.020705968141555786], "Conv_22/WithoutBiases/fq_input_0": [0.0, 0.00581572949886322], "Conv_129/WithoutBiases/fq_input_0": [0.0, 0.004710772540420294], "Conv_123/WithoutBiases/fq_input_0": [0.0, 0.016872208565473557], "Add_85/fq_input_0": [-0.02208520844578743, 0.02191266603767872], "Add_85/fq_input_1": [-0.012997213751077652, 0.012895672582089901], "Add_143/fq_input_0": [-0.015055370517075062, 0.01493775099515915], "Add_143/fq_input_1": [-0.008267931640148163, 0.008203338831663132], "Add_46/fq_input_0": [-0.025736436247825623, 0.02553536929190159], "Add_46/fq_input_1": [-0.025809651240706444, 0.025608014315366745], "Conv_96/WithoutBiases/fq_input_0": [-0.030720945447683334, 0.030480939894914627], "Conv_41/WithoutBiases/fq_input_0": [0.0, 0.004026732407510281], "Conv_100/WithoutBiases/fq_input_0": [0.0, 0.003673217026516795], "Conv_158/WithoutBiases/fq_input_0": [0.0, 0.0006734441849403083], "Add_27/fq_input_0": [-0.026998046785593033, 0.026787124574184418], "Add_27/fq_input_1": [-0.04204697161912918, 0.04171847924590111], "Conv_4/WithoutBiases/fq_input_0": [0.0, 0.007917627692222595], "Reshape_173/fq_input_0": [0.0, 0.03341957554221153]}

2
tools/pot/tests/data/reference_scale/mobilenet-v2-pytorch_weights.json
View File

File diff suppressed because one or more lines are too long

60
tools/pot/tests/data/reference_unify/mobilenet-v2-pytorch_to_unify.json
View File

@ -1,92 +1,92 @@
[ [
[ [
[ [
"351" "Add_114"
], ],
[ [
"351/fq_input_0", "Add_114/fq_input_0",
"351/fq_input_1" "Add_114/fq_input_1"
] ]
], ],
[ [
[ [
"376" "Add_124"
], ],
[ [
"376/fq_input_0", "Add_124/fq_input_0",
"376/fq_input_1" "Add_124/fq_input_1"
] ]
], ],
[ [
[ [
"389" "Add_143"
], ],
[ [
"389/fq_input_0", "Add_143/fq_input_0",
"389/fq_input_1" "Add_143/fq_input_1"
] ]
], ],
[ [
[ [
"414" "Add_153"
], ],
[ [
"414/fq_input_0", "Add_153/fq_input_0",
"414/fq_input_1" "Add_153/fq_input_1"
] ]
], ],
[ [
[ [
"427" "Add_27"
], ],
[ [
"427/fq_input_0", "Add_27/fq_input_0",
"427/fq_input_1" "Add_27/fq_input_1"
] ]
], ],
[ [
[ [
"440" "Add_46"
], ],
[ [
"440/fq_input_0", "Add_46/fq_input_0",
"440/fq_input_1" "Add_46/fq_input_1"
] ]
], ],
[ [
[ [
"465" "Add_56"
], ],
[ [
"465/fq_input_0", "Add_56/fq_input_0",
"465/fq_input_1" "Add_56/fq_input_1"
] ]
], ],
[ [
[ [
"478" "Add_75"
], ],
[ [
"478/fq_input_0", "Add_75/fq_input_0",
"478/fq_input_1" "Add_75/fq_input_1"
] ]
], ],
[ [
[ [
"503" "Add_85"
], ],
[ [
"503/fq_input_0", "Add_85/fq_input_0",
"503/fq_input_1" "Add_85/fq_input_1"
] ]
], ],
[ [
[ [
"516" "Add_95"
], ],
[ [
"516/fq_input_0", "Add_95/fq_input_0",
"516/fq_input_1" "Add_95/fq_input_1"
] ]
] ]
] ]

4
tools/pot/tests/data/test_cases_refs/mobilenet-v2-pytorch_statistics_data.json
View File

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1 version https://git-lfs.github.com/spec/v1
oid sha256:58579dc94614d990c0a13dbe189af561ca0a6a576261096bf2a70fb0cea1221c oid sha256:cbc6a94dc5d3e64fcc1bfe7a5f2c484a0f2fd8aa17fa9bfa8e013babf9c1903e
size 3423642 size 3425610

44
tools/pot/tests/test_graph.py
View File

@ -93,9 +93,9 @@ def test_build_quantization_graph_with_ignored_params(
} }
if model_name == 'resnet_example': if model_name == 'resnet_example':
ignored_params['scope'] = ['Convolution_283', 'Convolution_724'] ignored_params['scope'] = ['Conv_11/WithoutBiases', 'Conv_29/WithoutBiases']
elif model_name == 'googlenet_example': elif model_name == 'googlenet_example':
node_name = 'Convolution_289' node_name = 'Conv_10/WithoutBiases'
ignored_params['scope'] = [node_name] ignored_params['scope'] = [node_name]
elif model_name == 'mtcnn': elif model_name == 'mtcnn':
ignored_params = { ignored_params = {
@ -159,19 +159,19 @@ def test_build_quantization_graph_with_ignored_agnostic_params(
TEST_MODELS_REMOVAL = [ TEST_MODELS_REMOVAL = [
('mobilenetv2_ssd_example', 'pytorch', ['Convolution_448', ('mobilenetv2_ssd_example', 'pytorch', ['Conv_12/WithoutBiases',
'GroupConvolution_840', 'Conv_26/WithoutBiases',
'Convolution_1160']), 'Conv_41/WithoutBiases']),
('squeezenet1_1_example', 'pytorch', ['Convolution_150', ('squeezenet1_1_example', 'pytorch', ['Conv_5/WithoutBiases',
'Convolution_991']), 'Conv_47/WithoutBiases']),
('mobilenetv2_example', 'pytorch', ['Convolution_521', ('mobilenetv2_example', 'pytorch', ['Conv_10/WithoutBiases',
'GroupConvolution_863', 'Conv_18/WithoutBiases',
'Convolution_2450']), 'Conv_60/WithoutBiases']),
('googlenet_example', 'pytorch', ['Convolution_190', ('googlenet_example', 'pytorch', ['Conv_5/WithoutBiases',
'Convolution_289', 'Conv_10/WithoutBiases',
'Convolution_486', 'Conv_19/WithoutBiases',
'Convolution_1917', 'Conv_87/WithoutBiases',
'Convolution_2017']), 'Conv_93/WithoutBiases']),
('multiple_out_ports_net', 'tf', ['add_indices']) ('multiple_out_ports_net', 'tf', ['add_indices'])
] ]
@ -234,7 +234,8 @@ TEST_MODELS_WITHOUT_FQ_MOVING = [
ids=['{}_{}'.format(m[0], m[1]) for m in TEST_MODELS_WITHOUT_FQ_MOVING]) ids=['{}_{}'.format(m[0], m[1]) for m in TEST_MODELS_WITHOUT_FQ_MOVING])
def test_multibranch_propagation_without_fq_moving(tmp_path, models, model_name, model_framework): def test_multibranch_propagation_without_fq_moving(tmp_path, models, model_name, model_framework):
ignored_params = { ignored_params = {
"scope": ['Convolution_104', 'Convolution_152', 'Convolution_8', 'Convolution_56'] # Ignoring quantization for the first 4 convolution in the model
"scope": ['8/WithoutBiases', '9/WithoutBiases', '10/WithoutBiases', '11/WithoutBiases']
} }
model = models.get(model_name, model_framework, tmp_path) model = models.get(model_name, model_framework, tmp_path)
@ -243,9 +244,11 @@ def test_multibranch_propagation_without_fq_moving(tmp_path, models, model_name,
hardware_config = HardwareConfig.from_json((HARDWARE_CONFIG_PATH / 'cpu.json').as_posix()) hardware_config = HardwareConfig.from_json((HARDWARE_CONFIG_PATH / 'cpu.json').as_posix())
quantized_model = GraphTransformer(hardware_config).insert_fake_quantize(model, ignored_params) quantized_model = GraphTransformer(hardware_config).insert_fake_quantize(model, ignored_params)
node = get_node_by_name(quantized_model, 'Convolution_201') # Checking last convolution has FQ at inputs
node = get_node_by_name(quantized_model, '13/WithoutBiases')
for node_input in get_node_inputs(node)[:2]: for node_input in get_node_inputs(node)[:2]:
assert node_input.type == 'FakeQuantize' assert node_input.type == 'FakeQuantize'
# Checking ignored convolutions has no quantizers on inputs
assert len(get_nodes_by_type(quantized_model, ['FakeQuantize'])) == 2 assert len(get_nodes_by_type(quantized_model, ['FakeQuantize'])) == 2
@ -286,7 +289,8 @@ TEST_MODELS_WITHOUT_FQ_MOVING = [
ids=['{}_{}'.format(m[0], m[1]) for m in TEST_MODELS_WITHOUT_FQ_MOVING]) ids=['{}_{}'.format(m[0], m[1]) for m in TEST_MODELS_WITHOUT_FQ_MOVING])
def test_multibranch_propagation_with_fq_moving(tmp_path, models, model_name, model_framework): def test_multibranch_propagation_with_fq_moving(tmp_path, models, model_name, model_framework):
ignored_params = { ignored_params = {
"scope": ['Convolution_104', 'Convolution_152', 'Convolution_8', 'Convolution_56'] # Ignoring quantization for the first 4 convolution in the model
"scope": ['8/WithoutBiases', '9/WithoutBiases', '10/WithoutBiases', '11/WithoutBiases']
} }
model = models.get(model_name, model_framework, tmp_path) model = models.get(model_name, model_framework, tmp_path)
@ -308,8 +312,8 @@ def test_multibranch_propagation_with_fq_moving(tmp_path, models, model_name, mo
MODELS_FOR_FIRST_CONV_TEST = [ MODELS_FOR_FIRST_CONV_TEST = [
('1_input_model', 'onnx', ['Convolution_19']), ('1_input_model', 'onnx', ['Conv_3/WithoutBiases']),
('3_inputs_model', 'onnx', ['Convolution_172', 'Convolution_123', 'Convolution_27']), ('3_inputs_model', 'onnx', ['Conv_3/WithoutBiases', 'Conv_5/WithoutBiases', 'Conv_7/WithoutBiases']),
] ]

20
tools/pot/tests/test_sanity.py
View File

@ -23,26 +23,26 @@ from .utils.config import get_engine_config, merge_configs, \
get_dataset_info, PATHS2DATASETS_CONFIG, make_algo_config get_dataset_info, PATHS2DATASETS_CONFIG, make_algo_config
TEST_MODELS = [ TEST_MODELS = [
('mobilenet-v2-pytorch', 'pytorch', 'DefaultQuantization', 'performance', 300, {'accuracy@top1': 0.737, ('mobilenet-v2-pytorch', 'pytorch', 'DefaultQuantization', 'performance', 300, {'accuracy@top1': 0.729,
'accuracy@top5': 0.909}, 'accuracy@top5': 0.907},
{}, 'CPU'), {}, 'CPU'),
('mobilenet-v2-pytorch', 'pytorch', 'DefaultQuantization', 'mixed', 300, {'accuracy@top1': 0.731, ('mobilenet-v2-pytorch', 'pytorch', 'DefaultQuantization', 'mixed', 300, {'accuracy@top1': 0.734,
'accuracy@top5': 0.908}, 'accuracy@top5': 0.91},
{}, 'CPU'), {}, 'CPU'),
('mobilenet-v1-1.0-224-tf', 'tf', 'DefaultQuantization', 'performance', 100, {'accuracy@top1': 0.728, ('mobilenet-v1-1.0-224-tf', 'tf', 'DefaultQuantization', 'performance', 100, {'accuracy@top1': 0.728,
'accuracy@top5': 0.909}, 'accuracy@top5': 0.909},
{'use_fast_bias': False}, 'CPU'), {'use_fast_bias': False}, 'CPU'),
('mobilenet-v1-1.0-224-tf', 'tf', 'DefaultQuantization', 'performance', 100, {'accuracy@top1': 0.728, ('mobilenet-v1-1.0-224-tf', 'tf', 'DefaultQuantization', 'performance', 100, {'accuracy@top1': 0.723,
'accuracy@top5': 0.911}, 'accuracy@top5': 0.907},
{}, 'CPU'), {}, 'CPU'),
('mobilenet-ssd', 'caffe', 'AccuracyAwareQuantization', 'performance', 300, {'map': 0.6801}, ('mobilenet-ssd', 'caffe', 'AccuracyAwareQuantization', 'performance', 300, {'map': 0.6741},
{'metric_subset_ratio': 1.0, 'max_iter_num': 1, 'metrics': [{'name': 'map', 'baseline_value': 0.669}]}, 'CPU'), {'metric_subset_ratio': 1.0, 'max_iter_num': 1, 'metrics': [{'name': 'map', 'baseline_value': 0.669}]}, 'CPU'),
('mobilenet-ssd', 'caffe', 'AccuracyAwareQuantization', 'performance', 300, {'map': 0.6801}, ('mobilenet-ssd', 'caffe', 'AccuracyAwareQuantization', 'performance', 300, {'map': 0.6741},
{'metric_subset_ratio': 1.0, 'max_iter_num': 1, 'tune_hyperparams': True, {'metric_subset_ratio': 1.0, 'max_iter_num': 1, 'tune_hyperparams': True,
'metrics': [{'name': 'map', 'baseline_value': 0.669}]}, 'CPU'), 'metrics': [{'name': 'map', 'baseline_value': 0.669}]}, 'CPU'),
@ -53,7 +53,7 @@ TEST_MODELS = [
('mtcnn', 'caffe', 'DefaultQuantization', 'performance', 1, {'recall': 0.76, 'map': 0.6618}, {}, 'CPU'), ('mtcnn', 'caffe', 'DefaultQuantization', 'performance', 1, {'recall': 0.76, 'map': 0.6618}, {}, 'CPU'),
('mtcnn', 'caffe', 'DefaultQuantization', 'performance', 2, {'recall': 0.68, 'map': 0.4406}, ('mtcnn', 'caffe', 'DefaultQuantization', 'performance', 2, {'recall': 0.76, 'map': 0.5101},
{'use_fast_bias': False}, 'CPU'), {'use_fast_bias': False}, 'CPU'),
('octave-resnet-26-0.25', 'mxnet', 'DefaultQuantization', 'performance', 300, ('octave-resnet-26-0.25', 'mxnet', 'DefaultQuantization', 'performance', 300,
{'accuracy@top1': 0.766, 'accuracy@top5': 0.927}, {'use_fast_bias': False}, 'CPU'), {'accuracy@top1': 0.766, 'accuracy@top5': 0.927}, {'use_fast_bias': False}, 'CPU'),
@ -173,7 +173,7 @@ SIMPLIFIED_TEST_MODELS = [
('mobilenet-v2-pytorch', 'pytorch', 'DefaultQuantization', 'performance', ('mobilenet-v2-pytorch', 'pytorch', 'DefaultQuantization', 'performance',
{'accuracy@top1': 0.701, 'accuracy@top5': 0.91}, []), {'accuracy@top1': 0.701, 'accuracy@top5': 0.91}, []),
('mobilenet-v2-pytorch', 'pytorch', 'DefaultQuantization', 'performance', ('mobilenet-v2-pytorch', 'pytorch', 'DefaultQuantization', 'performance',
{'accuracy@top1': 0.712, 'accuracy@top5': 0.906}, ['--input_shape=[1,3,?,?]']) {'accuracy@top1': 0.703, 'accuracy@top5': 0.904}, ['--input_shape=[1,3,?,?]'])
] ]

16
tools/pot/tests/test_scales.py
View File

@ -16,6 +16,9 @@ from openvino.tools.pot.graph.node_utils import get_node_inputs, get_node_input,
from openvino.tools.pot.graph import model_utils as mu from openvino.tools.pot.graph import model_utils as mu
from openvino.tools.pot.statistics.collector import StatisticsCollector from openvino.tools.pot.statistics.collector import StatisticsCollector
EPS = 1e-6
class NumpyEncoder(json.JSONEncoder): class NumpyEncoder(json.JSONEncoder):
""" Special json encoder for numpy types """ """ Special json encoder for numpy types """
# pylint: disable=W0221, E0202 # pylint: disable=W0221, E0202
@ -28,6 +31,7 @@ class NumpyEncoder(json.JSONEncoder):
return o.tolist() return o.tolist()
return json.JSONEncoder.default(self, o) return json.JSONEncoder.default(self, o)
def get_fq_nodes_stats_algo(model, preset, bits, is_weights, clipping_value=None): def get_fq_nodes_stats_algo(model, preset, bits, is_weights, clipping_value=None):
test_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), test_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)),
'./data/reference_scale/test_data') './data/reference_scale/test_data')
@ -129,7 +133,6 @@ def test_activation_scales(tmp_path, models, preset, bits, stats_path, clipping_
assert len(ref_nodes) == len(nodes) assert len(ref_nodes) == len(nodes)
processed_nodes = [] processed_nodes = []
eps = 1e-3
for ref_name in ref_nodes: for ref_name in ref_nodes:
ref_min, ref_max = ref_nodes[ref_name] ref_min, ref_max = ref_nodes[ref_name]
@ -137,7 +140,7 @@ def test_activation_scales(tmp_path, models, preset, bits, stats_path, clipping_
error = max(abs(node_min - ref_min), abs(node_max - ref_max)) error = max(abs(node_min - ref_min), abs(node_max - ref_max))
if error <= eps and ref_name not in processed_nodes: if error <= EPS and ref_name not in processed_nodes:
processed_nodes.append(( processed_nodes.append((
ref_name, abs(node_min - ref_min), abs(node_max - ref_max))) ref_name, abs(node_min - ref_min), abs(node_max - ref_max)))
@ -163,8 +166,8 @@ def test_weights_scales(tmp_path, models):
ref_min, ref_max = ref_weights[fq_name]['low_level'], ref_weights[fq_name]['high_level'] ref_min, ref_max = ref_weights[fq_name]['low_level'], ref_weights[fq_name]['high_level']
assert_flag = False assert_flag = False
if np.array_equal(ref_min, item_min) and \ if np.max(np.abs(ref_min - item_min)) < EPS and \
np.array_equal(ref_max, item_max): np.max(np.abs(ref_max - item_max)) < EPS:
assert_flag = True assert_flag = True
if not assert_flag: if not assert_flag:
@ -259,7 +262,6 @@ def test_fake_quantize_configurations(tmp_path, models, model_name, model_framew
local_file = open(local_path, 'w') local_file = open(local_path, 'w')
model_values = {} model_values = {}
eps = 1e-3
fq_list = mu.get_nodes_by_type(model, ['FakeQuantize']) fq_list = mu.get_nodes_by_type(model, ['FakeQuantize'])
for fq in sorted(fq_list, key=lambda item: item.name): for fq in sorted(fq_list, key=lambda item: item.name):
min_levels, max_levels = tuple([get_node_value(node) min_levels, max_levels = tuple([get_node_value(node)
@ -292,8 +294,8 @@ def test_fake_quantize_configurations(tmp_path, models, model_name, model_framew
for min_level, max_level, ref_min, ref_max in zip( for min_level, max_level, ref_min, ref_max in zip(
min_levels, max_levels, refs_min_levels, refs_max_levels): min_levels, max_levels, refs_min_levels, refs_max_levels):
assert abs(min_level - ref_min) < eps assert abs(min_level - ref_min) < EPS
assert abs(max_level - ref_max) < eps assert abs(max_level - ref_max) < EPS
def _get_pytorch_accuracy_checker_config(path_to_dataset): def _get_pytorch_accuracy_checker_config(path_to_dataset):

2
tools/pot/tests/test_statistics_collector.py
View File

@ -63,7 +63,7 @@ def test_statistics_collector_subsets(tmp_path, models, model_name, model_framew
with open(refs_file.as_posix()) as file: with open(refs_file.as_posix()) as file:
refs = json.load(file) refs = json.load(file)
eps = 1e-3 eps = 1e-6
local_out = {} local_out = {}
for algo_name, algo_val in out.items(): for algo_name, algo_val in out.items():
local_out[algo_name] = {} local_out[algo_name] = {}