Open mingxiaoh opened 5 years ago
Hello, we tried to reproduce the ResNet50 inference, but only got 73+% top1, it is still far from SOTA(75%) and your data(76%). Any comments here?
Details: torch (1.1.0a0+a3933b8) torchvision (0.2.2.post3)
(py3-intel-chainer) [mingxiao@mlt-gpu207 examples]$ python imagenet_eval.py --data ~/data/ -a resnet50 -b 20 -e => creating model 'resnet50' => using pre-trained parameters 'imagenet' Images transformed from size 256 to [3, 224, 224] Test: [0/2500] Time 2.704 (2.704) Loss 0.7271 (0.7271) Acc@1 85.000 (85.000) Acc@5 90.000 (90.000) Test: [10/2500] Time 0.037 (0.279) Loss 0.7189 (0.5648) Acc@1 75.000 (82.727) Acc@5 95.000 (96.364) Test: [20/2500] Time 0.036 (0.164) Loss 0.9403 (0.7057) Acc@1 80.000 (80.714) Acc@5 90.000 (94.762) Test: [30/2500] Time 0.036 (0.123) Loss 0.5037 (0.6162) Acc@1 90.000 (83.871) Acc@5 95.000 (95.645) Test: [40/2500] Time 0.036 (0.102) Loss 0.7368 (0.5309) Acc@1 85.000 (85.976) Acc@5 85.000 (95.854) Test: [50/2500] Time 0.036 (0.089) Loss 0.6895 (0.4964) Acc@1 80.000 (86.961) Acc@5 90.000 (96.078) Test: [60/2500] Time 0.036 (0.080) Loss 0.7518 (0.5163) Acc@1 90.000 (86.967) Acc@5 95.000 (95.820) Test: [70/2500] Time 0.036 (0.074) Loss 0.2763 (0.5536) Acc@1 90.000 (85.845) Acc@5 100.000 (95.986) Test: [80/2500] Time 0.037 (0.069) Loss 2.0607 (0.6061) Acc@1 40.000 (85.185) Acc@5 90.000 (95.370) Test: [90/2500] Time 0.036 (0.066) Loss 1.9135 (0.7201) Acc@1 40.000 (82.253) Acc@5 95.000 (94.176) Test: [100/2500] Time 0.036 (0.063) Loss 0.7474 (0.7224) Acc@1 50.000 (81.634) Acc@5 100.000 (94.505) Test: [110/2500] Time 0.036 (0.060) Loss 1.3561 (0.7373) Acc@1 65.000 (80.946) Acc@5 95.000 (94.459) Test: [120/2500] Time 0.037 (0.059) Loss 0.4951 (0.7848) Acc@1 90.000 (79.959) Acc@5 100.000 (94.008) Test: [130/2500] Time 0.037 (0.057) Loss 0.5267 (0.7856) Acc@1 80.000 (79.924) Acc@5 100.000 (93.893) Test: [140/2500] Time 0.036 (0.055) Loss 0.8788 (0.8200) Acc@1 75.000 (79.291) Acc@5 95.000 (93.617) Test: [150/2500] Time 0.036 (0.054) Loss 2.6851 (0.8557) Acc@1 35.000 (78.543) Acc@5 85.000 (93.245) Test: [160/2500] Time 0.036 (0.053) Loss 2.0516 (0.8964) Acc@1 50.000 (77.422) Acc@5 90.000 (93.043) Test: [170/2500] Time 0.036 (0.052) Loss 0.9259 (0.9198) Acc@1 55.000 (76.579) Acc@5 100.000 (92.865) Test: [180/2500] Time 0.036 (0.051) Loss 0.6522 (0.9092) Acc@1 75.000 (76.878) Acc@5 100.000 (92.956) Test: [190/2500] Time 0.036 (0.050) Loss 0.2513 (0.9095) Acc@1 85.000 (76.230) Acc@5 100.000 (93.141) Test: [200/2500] Time 0.037 (0.050) Loss 0.2398 (0.9087) Acc@1 95.000 (76.567) Acc@5 100.000 (93.159) Test: [210/2500] Time 0.037 (0.049) Loss 0.3800 (0.8916) Acc@1 90.000 (77.038) Acc@5 95.000 (93.294) Test: [220/2500] Time 0.036 (0.049) Loss 0.6269 (0.8736) Acc@1 90.000 (77.511) Acc@5 95.000 (93.462) Test: [230/2500] Time 0.036 (0.048) Loss 0.0200 (0.8483) Acc@1 100.000 (78.225) Acc@5 100.000 (93.615) Test: [240/2500] Time 0.036 (0.048) Loss 0.0885 (0.8253) Acc@1 95.000 (78.900) Acc@5 100.000 (93.797) Test: [250/2500] Time 0.037 (0.047) Loss 0.0069 (0.8111) Acc@1 100.000 (79.283) Acc@5 100.000 (93.944) Test: [260/2500] Time 0.036 (0.047) Loss 0.3296 (0.8053) Acc@1 95.000 (79.368) Acc@5 95.000 (93.946) Test: [270/2500] Time 0.036 (0.046) Loss 1.4653 (0.8146) Acc@1 60.000 (79.299) Acc@5 90.000 (93.875) Test: [280/2500] Time 0.036 (0.046) Loss 1.3985 (0.8086) Acc@1 65.000 (79.395) Acc@5 80.000 (93.932) Test: [290/2500] Time 0.037 (0.046) Loss 0.3668 (0.8072) Acc@1 90.000 (79.519) Acc@5 100.000 (93.883) Test: [300/2500] Time 0.037 (0.045) Loss 0.7728 (0.8094) Acc@1 80.000 (79.535) Acc@5 95.000 (93.920) Test: [310/2500] Time 0.037 (0.045) Loss 1.3635 (0.8137) Acc@1 65.000 (79.357) Acc@5 95.000 (93.955) Test: [320/2500] Time 0.036 (0.045) Loss 0.8331 (0.8190) Acc@1 85.000 (79.346) Acc@5 95.000 (93.941) Test: [330/2500] Time 0.036 (0.045) Loss 0.4665 (0.8053) Acc@1 90.000 (79.653) Acc@5 95.000 (94.033) Test: [340/2500] Time 0.037 (0.044) Loss 0.0864 (0.7940) Acc@1 95.000 (79.941) Acc@5 100.000 (94.120) Test: [350/2500] Time 0.036 (0.044) Loss 1.1962 (0.7807) Acc@1 75.000 (80.271) Acc@5 95.000 (94.231) Test: [360/2500] Time 0.036 (0.044) Loss 0.2139 (0.7706) Acc@1 95.000 (80.512) Acc@5 95.000 (94.294) Test: [370/2500] Time 0.037 (0.044) Loss 0.1379 (0.7592) Acc@1 95.000 (80.809) Acc@5 100.000 (94.394) Test: [380/2500] Time 0.036 (0.043) Loss 1.0197 (0.7602) Acc@1 75.000 (80.866) Acc@5 90.000 (94.370) Test: [390/2500] Time 0.037 (0.043) Loss 0.1889 (0.7601) Acc@1 95.000 (80.895) Acc@5 100.000 (94.412) Test: [400/2500] Time 0.036 (0.043) Loss 0.2492 (0.7598) Acc@1 90.000 (80.885) Acc@5 100.000 (94.439) Test: [410/2500] Time 0.037 (0.043) Loss 0.2367 (0.7663) Acc@1 95.000 (80.779) Acc@5 100.000 (94.367) Test: [420/2500] Time 0.036 (0.043) Loss 1.1600 (0.7876) Acc@1 60.000 (80.036) Acc@5 85.000 (94.252) Test: [430/2500] Time 0.036 (0.043) Loss 1.1108 (0.7932) Acc@1 75.000 (79.988) Acc@5 90.000 (94.223) Test: [440/2500] Time 0.037 (0.043) Loss 0.5544 (0.7979) Acc@1 80.000 (79.796) Acc@5 100.000 (94.263) Test: [450/2500] Time 0.036 (0.042) Loss 1.0699 (0.7975) Acc@1 55.000 (79.723) Acc@5 95.000 (94.290) Test: [460/2500] Time 0.037 (0.042) Loss 1.0099 (0.7941) Acc@1 80.000 (79.783) Acc@5 90.000 (94.328) Test: [470/2500] Time 0.036 (0.042) Loss 0.6814 (0.8014) Acc@1 80.000 (79.544) Acc@5 100.000 (94.299) Test: [480/2500] Time 0.036 (0.042) Loss 0.5434 (0.8016) Acc@1 80.000 (79.511) Acc@5 100.000 (94.335) Test: [490/2500] Time 0.037 (0.042) Loss 0.7770 (0.8022) Acc@1 75.000 (79.430) Acc@5 95.000 (94.358) Test: [500/2500] Time 0.036 (0.042) Loss 1.1530 (0.8020) Acc@1 65.000 (79.401) Acc@5 100.000 (94.431) Test: [510/2500] Time 0.036 (0.042) Loss 0.9519 (0.8027) Acc@1 65.000 (79.247) Acc@5 95.000 (94.472) Test: [520/2500] Time 0.036 (0.042) Loss 0.2959 (0.8030) Acc@1 95.000 (79.165) Acc@5 100.000 (94.491) Test: [530/2500] Time 0.036 (0.042) Loss 0.8767 (0.8013) Acc@1 75.000 (79.134) Acc@5 100.000 (94.557) Test: [540/2500] Time 0.105 (0.042) Loss 0.3152 (0.8015) Acc@1 95.000 (79.168) Acc@5 100.000 (94.538) Test: [550/2500] Time 0.036 (0.042) Loss 1.6377 (0.7974) Acc@1 65.000 (79.274) Acc@5 85.000 (94.583) Test: [560/2500] Time 0.036 (0.042) Loss 0.4664 (0.7982) Acc@1 85.000 (79.269) Acc@5 95.000 (94.563) Test: [570/2500] Time 0.037 (0.042) Loss 0.6717 (0.8058) Acc@1 85.000 (79.177) Acc@5 95.000 (94.483) Test: [580/2500] Time 0.036 (0.042) Loss 0.4850 (0.8037) Acc@1 90.000 (79.157) Acc@5 100.000 (94.527) Test: [590/2500] Time 0.036 (0.042) Loss 0.7717 (0.8043) Acc@1 85.000 (79.120) Acc@5 95.000 (94.552) Test: [600/2500] Time 0.037 (0.042) Loss 2.1893 (0.8087) Acc@1 25.000 (78.993) Acc@5 100.000 (94.567) Test: [610/2500] Time 0.036 (0.041) Loss 0.0302 (0.8118) Acc@1 100.000 (78.756) Acc@5 100.000 (94.615) Test: [620/2500] Time 0.037 (0.041) Loss 1.2387 (0.8152) Acc@1 50.000 (78.647) Acc@5 95.000 (94.589) Test: [630/2500] Time 0.036 (0.041) Loss 0.5377 (0.8129) Acc@1 85.000 (78.558) Acc@5 95.000 (94.651) Test: [640/2500] Time 0.037 (0.041) Loss 0.4386 (0.8061) Acc@1 80.000 (78.736) Acc@5 100.000 (94.696) Test: [650/2500] Time 0.037 (0.041) Loss 0.5441 (0.8006) Acc@1 80.000 (78.902) Acc@5 100.000 (94.739) Test: [660/2500] Time 0.037 (0.041) Loss 0.7073 (0.7952) Acc@1 75.000 (79.032) Acc@5 100.000 (94.796) Test: [670/2500] Time 0.036 (0.041) Loss 0.3843 (0.7975) Acc@1 90.000 (78.815) Acc@5 95.000 (94.806) Test: [680/2500] Time 0.036 (0.041) Loss 2.9731 (0.8027) Acc@1 55.000 (78.744) Acc@5 90.000 (94.809) Test: [690/2500] Time 0.036 (0.041) Loss 0.7158 (0.8016) Acc@1 85.000 (78.828) Acc@5 95.000 (94.797) Test: [700/2500] Time 0.036 (0.041) Loss 0.8419 (0.8031) Acc@1 70.000 (78.738) Acc@5 95.000 (94.807) Test: [710/2500] Time 0.036 (0.041) Loss 0.0556 (0.8060) Acc@1 100.000 (78.579) Acc@5 100.000 (94.824) Test: [720/2500] Time 0.036 (0.041) Loss 0.8053 (0.8107) Acc@1 80.000 (78.530) Acc@5 100.000 (94.785) Test: [730/2500] Time 0.036 (0.041) Loss 0.7175 (0.8074) Acc@1 90.000 (78.618) Acc@5 100.000 (94.815) Test: [740/2500] Time 0.036 (0.041) Loss 0.6575 (0.8006) Acc@1 95.000 (78.806) Acc@5 95.000 (94.872) Test: [750/2500] Time 0.036 (0.041) Loss 0.2692 (0.8036) Acc@1 95.000 (78.815) Acc@5 95.000 (94.854) Test: [760/2500] Time 0.036 (0.040) Loss 0.9070 (0.8051) Acc@1 70.000 (78.739) Acc@5 100.000 (94.862) Test: [770/2500] Time 0.036 (0.040) Loss 0.5024 (0.8015) Acc@1 95.000 (78.800) Acc@5 95.000 (94.890) Test: [780/2500] Time 0.036 (0.040) Loss 1.1427 (0.8049) Acc@1 65.000 (78.713) Acc@5 95.000 (94.898) Test: [790/2500] Time 0.036 (0.040) Loss 0.7711 (0.8105) Acc@1 80.000 (78.603) Acc@5 95.000 (94.848) Test: [800/2500] Time 0.036 (0.040) Loss 0.2140 (0.8079) Acc@1 90.000 (78.658) Acc@5 100.000 (94.881) Test: [810/2500] Time 0.036 (0.040) Loss 0.1252 (0.8029) Acc@1 95.000 (78.798) Acc@5 100.000 (94.920) Test: [820/2500] Time 0.036 (0.040) Loss 0.4773 (0.7979) Acc@1 90.000 (78.959) Acc@5 95.000 (94.945) Test: [830/2500] Time 0.036 (0.040) Loss 0.0433 (0.7976) Acc@1 100.000 (78.995) Acc@5 100.000 (94.958) Test: [840/2500] Time 0.036 (0.040) Loss 0.6456 (0.7937) Acc@1 85.000 (79.132) Acc@5 95.000 (94.982) Test: [850/2500] Time 0.036 (0.040) Loss 0.3178 (0.7914) Acc@1 95.000 (79.236) Acc@5 95.000 (95.000) Test: [860/2500] Time 0.036 (0.040) Loss 0.7505 (0.7906) Acc@1 80.000 (79.228) Acc@5 90.000 (95.000) Test: [870/2500] Time 0.036 (0.040) Loss 0.5531 (0.7903) Acc@1 75.000 (79.214) Acc@5 100.000 (95.023) Test: [880/2500] Time 0.036 (0.040) Loss 0.3595 (0.7870) Acc@1 95.000 (79.257) Acc@5 100.000 (95.051) Test: [890/2500] Time 0.036 (0.040) Loss 1.7804 (0.7870) Acc@1 55.000 (79.237) Acc@5 95.000 (95.056) Test: [900/2500] Time 0.036 (0.040) Loss 0.8308 (0.7935) Acc@1 80.000 (79.057) Acc@5 90.000 (95.067) Test: [910/2500] Time 0.036 (0.040) Loss 0.0886 (0.7898) Acc@1 95.000 (79.166) Acc@5 100.000 (95.082) Test: [920/2500] Time 0.036 (0.040) Loss 1.3319 (0.7896) Acc@1 70.000 (79.224) Acc@5 95.000 (95.060) Test: [930/2500] Time 0.036 (0.040) Loss 0.0865 (0.7928) Acc@1 100.000 (79.216) Acc@5 100.000 (95.038) Test: [940/2500] Time 0.036 (0.040) Loss 0.0196 (0.7941) Acc@1 100.000 (79.203) Acc@5 100.000 (95.011) Test: [950/2500] Time 0.036 (0.040) Loss 1.4047 (0.7955) Acc@1 65.000 (79.138) Acc@5 90.000 (94.979) Test: [960/2500] Time 0.036 (0.040) Loss 0.7717 (0.7998) Acc@1 85.000 (79.022) Acc@5 95.000 (94.927) Test: [970/2500] Time 0.036 (0.040) Loss 0.2456 (0.7992) Acc@1 95.000 (79.001) Acc@5 95.000 (94.938) Test: [980/2500] Time 0.036 (0.040) Loss 0.3763 (0.8003) Acc@1 95.000 (79.006) Acc@5 95.000 (94.903) Test: [990/2500] Time 0.036 (0.039) Loss 0.3058 (0.8009) Acc@1 90.000 (78.991) Acc@5 95.000 (94.879) Test: [1000/2500] Time 0.036 (0.039) Loss 1.5105 (0.8021) Acc@1 55.000 (78.996) Acc@5 95.000 (94.880) Test: [1010/2500] Time 0.036 (0.039) Loss 0.1445 (0.8050) Acc@1 100.000 (78.937) Acc@5 100.000 (94.832) Test: [1020/2500] Time 0.036 (0.039) Loss 1.2508 (0.8037) Acc@1 70.000 (78.962) Acc@5 95.000 (94.853) Test: [1030/2500] Time 0.036 (0.039) Loss 1.7219 (0.8058) Acc@1 65.000 (78.923) Acc@5 80.000 (94.830) Test: [1040/2500] Time 0.036 (0.039) Loss 0.2517 (0.8198) Acc@1 100.000 (78.612) Acc@5 100.000 (94.673) Test: [1050/2500] Time 0.036 (0.039) Loss 0.1276 (0.8206) Acc@1 100.000 (78.578) Acc@5 100.000 (94.676) Test: [1060/2500] Time 0.036 (0.039) Loss 0.9815 (0.8214) Acc@1 70.000 (78.530) Acc@5 100.000 (94.694) Test: [1070/2500] Time 0.036 (0.039) Loss 1.5380 (0.8224) Acc@1 65.000 (78.492) Acc@5 90.000 (94.664) Test: [1080/2500] Time 0.036 (0.039) Loss 0.5527 (0.8213) Acc@1 80.000 (78.511) Acc@5 100.000 (94.672) Test: [1090/2500] Time 0.036 (0.039) Loss 1.7449 (0.8259) Acc@1 65.000 (78.368) Acc@5 85.000 (94.615) Test: [1100/2500] Time 0.036 (0.039) Loss 0.8507 (0.8296) Acc@1 75.000 (78.333) Acc@5 95.000 (94.569) Test: [1110/2500] Time 0.036 (0.039) Loss 0.6621 (0.8307) Acc@1 85.000 (78.281) Acc@5 100.000 (94.568) Test: [1120/2500] Time 0.036 (0.039) Loss 0.5863 (0.8357) Acc@1 80.000 (78.158) Acc@5 90.000 (94.514) Test: [1130/2500] Time 0.036 (0.039) Loss 1.4977 (0.8335) Acc@1 70.000 (78.223) Acc@5 90.000 (94.527) Test: [1140/2500] Time 0.036 (0.039) Loss 2.4943 (0.8396) Acc@1 50.000 (78.124) Acc@5 75.000 (94.452) Test: [1150/2500] Time 0.036 (0.039) Loss 1.2823 (0.8440) Acc@1 70.000 (78.076) Acc@5 90.000 (94.387) Test: [1160/2500] Time 0.036 (0.039) Loss 1.7010 (0.8497) Acc@1 55.000 (77.929) Acc@5 85.000 (94.337) Test: [1170/2500] Time 0.036 (0.039) Loss 0.9460 (0.8530) Acc@1 70.000 (77.852) Acc@5 90.000 (94.270) Test: [1180/2500] Time 0.036 (0.039) Loss 0.1823 (0.8582) Acc@1 95.000 (77.786) Acc@5 100.000 (94.191) Test: [1190/2500] Time 0.036 (0.039) Loss 0.0599 (0.8573) Acc@1 95.000 (77.813) Acc@5 100.000 (94.190) Test: [1200/2500] Time 0.036 (0.039) Loss 0.6744 (0.8641) Acc@1 80.000 (77.664) Acc@5 90.000 (94.105) Test: [1210/2500] Time 0.036 (0.039) Loss 0.7390 (0.8671) Acc@1 65.000 (77.519) Acc@5 100.000 (94.075) Test: [1220/2500] Time 0.036 (0.039) Loss 1.6444 (0.8700) Acc@1 50.000 (77.416) Acc@5 100.000 (94.054) Test: [1230/2500] Time 0.036 (0.039) Loss 1.2105 (0.8737) Acc@1 70.000 (77.360) Acc@5 95.000 (94.025) Test: [1240/2500] Time 0.036 (0.039) Loss 0.5262 (0.8801) Acc@1 85.000 (77.236) Acc@5 90.000 (93.924) Test: [1250/2500] Time 0.036 (0.039) Loss 0.3565 (0.8854) Acc@1 85.000 (77.142) Acc@5 100.000 (93.861) Test: [1260/2500] Time 0.036 (0.039) Loss 1.7938 (0.8913) Acc@1 45.000 (77.006) Acc@5 80.000 (93.775) Test: [1270/2500] Time 0.036 (0.039) Loss 0.9489 (0.8963) Acc@1 80.000 (76.904) Acc@5 95.000 (93.729) Test: [1280/2500] Time 0.036 (0.039) Loss 0.2843 (0.8965) Acc@1 100.000 (76.874) Acc@5 100.000 (93.716) Test: [1290/2500] Time 0.036 (0.039) Loss 2.0598 (0.8986) Acc@1 45.000 (76.824) Acc@5 85.000 (93.718) Test: [1300/2500] Time 0.036 (0.039) Loss 0.5175 (0.9018) Acc@1 90.000 (76.791) Acc@5 100.000 (93.689) Test: [1310/2500] Time 0.077 (0.039) Loss 1.3497 (0.9046) Acc@1 40.000 (76.709) Acc@5 95.000 (93.677) Test: [1320/2500] Time 0.036 (0.039) Loss 0.9780 (0.9085) Acc@1 65.000 (76.578) Acc@5 85.000 (93.656) Test: [1330/2500] Time 0.036 (0.039) Loss 0.9246 (0.9132) Acc@1 70.000 (76.476) Acc@5 90.000 (93.621) Test: [1340/2500] Time 0.036 (0.039) Loss 2.1366 (0.9145) Acc@1 55.000 (76.454) Acc@5 85.000 (93.594) Test: [1350/2500] Time 0.036 (0.039) Loss 0.3810 (0.9145) Acc@1 90.000 (76.429) Acc@5 95.000 (93.601) Test: [1360/2500] Time 0.036 (0.039) Loss 0.5766 (0.9179) Acc@1 90.000 (76.356) Acc@5 100.000 (93.564) Test: [1370/2500] Time 0.036 (0.039) Loss 0.2055 (0.9168) Acc@1 95.000 (76.371) Acc@5 100.000 (93.567) Test: [1380/2500] Time 0.070 (0.039) Loss 0.2993 (0.9179) Acc@1 90.000 (76.336) Acc@5 95.000 (93.559) Test: [1390/2500] Time 0.036 (0.039) Loss 1.2469 (0.9162) Acc@1 70.000 (76.387) Acc@5 95.000 (93.562) Test: [1400/2500] Time 0.036 (0.039) Loss 0.2995 (0.9225) Acc@1 95.000 (76.299) Acc@5 95.000 (93.487) Test: [1410/2500] Time 0.036 (0.039) Loss 0.4661 (0.9195) Acc@1 85.000 (76.364) Acc@5 95.000 (93.505) Test: [1420/2500] Time 0.036 (0.039) Loss 0.3127 (0.9195) Acc@1 95.000 (76.383) Acc@5 100.000 (93.487) Test: [1430/2500] Time 0.036 (0.039) Loss 1.5971 (0.9186) Acc@1 60.000 (76.433) Acc@5 80.000 (93.484) Test: [1440/2500] Time 0.036 (0.039) Loss 0.0613 (0.9173) Acc@1 100.000 (76.492) Acc@5 100.000 (93.487) Test: [1450/2500] Time 0.036 (0.039) Loss 0.5770 (0.9166) Acc@1 80.000 (76.465) Acc@5 100.000 (93.501) Test: [1460/2500] Time 0.036 (0.039) Loss 2.6388 (0.9189) Acc@1 45.000 (76.451) Acc@5 55.000 (93.474) Test: [1470/2500] Time 0.036 (0.039) Loss 1.0036 (0.9248) Acc@1 85.000 (76.360) Acc@5 90.000 (93.402) Test: [1480/2500] Time 0.036 (0.039) Loss 0.8197 (0.9289) Acc@1 80.000 (76.283) Acc@5 95.000 (93.352) Test: [1490/2500] Time 0.036 (0.039) Loss 1.8379 (0.9277) Acc@1 60.000 (76.328) Acc@5 85.000 (93.357) Test: [1500/2500] Time 0.036 (0.039) Loss 2.8159 (0.9339) Acc@1 40.000 (76.219) Acc@5 75.000 (93.294) Test: [1510/2500] Time 0.036 (0.039) Loss 0.0468 (0.9372) Acc@1 100.000 (76.168) Acc@5 100.000 (93.250) Test: [1520/2500] Time 0.036 (0.039) Loss 1.1193 (0.9350) Acc@1 75.000 (76.239) Acc@5 95.000 (93.261) Test: [1530/2500] Time 0.036 (0.039) Loss 0.3309 (0.9345) Acc@1 95.000 (76.267) Acc@5 95.000 (93.246) Test: [1540/2500] Time 0.036 (0.039) Loss 1.8861 (0.9341) Acc@1 75.000 (76.298) Acc@5 80.000 (93.228) Test: [1550/2500] Time 0.036 (0.039) Loss 1.8906 (0.9383) Acc@1 35.000 (76.167) Acc@5 85.000 (93.182) Test: [1560/2500] Time 0.036 (0.039) Loss 1.3631 (0.9441) Acc@1 60.000 (76.019) Acc@5 90.000 (93.107) Test: [1570/2500] Time 0.036 (0.039) Loss 0.3538 (0.9437) Acc@1 85.000 (76.047) Acc@5 100.000 (93.090) Test: [1580/2500] Time 0.036 (0.039) Loss 1.6606 (0.9441) Acc@1 60.000 (76.031) Acc@5 80.000 (93.080) Test: [1590/2500] Time 0.036 (0.039) Loss 0.7779 (0.9503) Acc@1 80.000 (75.889) Acc@5 100.000 (93.011) Test: [1600/2500] Time 0.036 (0.039) Loss 0.2365 (0.9523) Acc@1 95.000 (75.731) Acc@5 100.000 (93.017) Test: [1610/2500] Time 0.036 (0.039) Loss 0.9663 (0.9524) Acc@1 70.000 (75.739) Acc@5 95.000 (93.014) Test: [1620/2500] Time 0.036 (0.039) Loss 1.4808 (0.9528) Acc@1 65.000 (75.762) Acc@5 90.000 (93.001) Test: [1630/2500] Time 0.036 (0.039) Loss 1.9637 (0.9574) Acc@1 60.000 (75.696) Acc@5 75.000 (92.934) Test: [1640/2500] Time 0.036 (0.039) Loss 2.0634 (0.9611) Acc@1 55.000 (75.637) Acc@5 90.000 (92.895) Test: [1650/2500] Time 0.036 (0.039) Loss 1.1165 (0.9642) Acc@1 75.000 (75.533) Acc@5 90.000 (92.856) Test: [1660/2500] Time 0.037 (0.039) Loss 2.1994 (0.9665) Acc@1 30.000 (75.458) Acc@5 80.000 (92.839) Test: [1670/2500] Time 0.037 (0.039) Loss 0.4767 (0.9670) Acc@1 85.000 (75.416) Acc@5 100.000 (92.849) Test: [1680/2500] Time 0.036 (0.039) Loss 0.3436 (0.9643) Acc@1 90.000 (75.467) Acc@5 95.000 (92.879) Test: [1690/2500] Time 0.036 (0.039) Loss 2.8868 (0.9682) Acc@1 45.000 (75.387) Acc@5 75.000 (92.850) Test: [1700/2500] Time 0.036 (0.039) Loss 0.9127 (0.9718) Acc@1 75.000 (75.373) Acc@5 95.000 (92.798) Test: [1710/2500] Time 0.036 (0.039) Loss 1.3394 (0.9741) Acc@1 70.000 (75.324) Acc@5 85.000 (92.779) Test: [1720/2500] Time 0.036 (0.039) Loss 0.0846 (0.9728) Acc@1 100.000 (75.369) Acc@5 100.000 (92.780) Test: [1730/2500] Time 0.036 (0.039) Loss 3.2437 (0.9778) Acc@1 30.000 (75.257) Acc@5 65.000 (92.733) Test: [1740/2500] Time 0.036 (0.039) Loss 2.0856 (0.9784) Acc@1 80.000 (75.253) Acc@5 85.000 (92.720) Test: [1750/2500] Time 0.036 (0.039) Loss 0.9843 (0.9790) Acc@1 75.000 (75.237) Acc@5 90.000 (92.704) Test: [1760/2500] Time 0.036 (0.039) Loss 0.6655 (0.9791) Acc@1 80.000 (75.233) Acc@5 95.000 (92.712) Test: [1770/2500] Time 0.036 (0.039) Loss 1.2545 (0.9809) Acc@1 65.000 (75.189) Acc@5 90.000 (92.699) Test: [1780/2500] Time 0.036 (0.039) Loss 1.7366 (0.9827) Acc@1 70.000 (75.143) Acc@5 85.000 (92.681) Test: [1790/2500] Time 0.036 (0.039) Loss 0.8464 (0.9822) Acc@1 95.000 (75.179) Acc@5 95.000 (92.683) Test: [1800/2500] Time 0.036 (0.039) Loss 0.7508 (0.9816) Acc@1 75.000 (75.183) Acc@5 100.000 (92.685) Test: [1810/2500] Time 0.036 (0.039) Loss 0.5370 (0.9809) Acc@1 80.000 (75.215) Acc@5 100.000 (92.681) Test: [1820/2500] Time 0.036 (0.039) Loss 3.9797 (0.9830) Acc@1 35.000 (75.181) Acc@5 60.000 (92.652) Test: [1830/2500] Time 0.036 (0.039) Loss 0.7985 (0.9887) Acc@1 80.000 (75.090) Acc@5 90.000 (92.567) Test: [1840/2500] Time 0.036 (0.039) Loss 1.1496 (0.9899) Acc@1 80.000 (75.062) Acc@5 85.000 (92.539) Test: [1850/2500] Time 0.036 (0.039) Loss 2.3159 (0.9894) Acc@1 55.000 (75.092) Acc@5 70.000 (92.545) Test: [1860/2500] Time 0.036 (0.039) Loss 1.4142 (0.9915) Acc@1 50.000 (75.032) Acc@5 90.000 (92.517) Test: [1870/2500] Time 0.036 (0.039) Loss 1.5630 (0.9932) Acc@1 50.000 (74.992) Acc@5 90.000 (92.496) Test: [1880/2500] Time 0.036 (0.039) Loss 1.0481 (0.9947) Acc@1 80.000 (74.942) Acc@5 85.000 (92.485) Test: [1890/2500] Time 0.041 (0.039) Loss 0.4099 (0.9943) Acc@1 90.000 (74.960) Acc@5 100.000 (92.491) Test: [1900/2500] Time 0.036 (0.039) Loss 1.0773 (0.9940) Acc@1 75.000 (74.961) Acc@5 95.000 (92.501) Test: [1910/2500] Time 0.036 (0.039) Loss 2.7580 (0.9963) Acc@1 25.000 (74.927) Acc@5 80.000 (92.475) Test: [1920/2500] Time 0.036 (0.039) Loss 0.9038 (0.9986) Acc@1 80.000 (74.901) Acc@5 90.000 (92.431) Test: [1930/2500] Time 0.036 (0.039) Loss 1.7654 (0.9984) Acc@1 45.000 (74.904) Acc@5 80.000 (92.431) Test: [1940/2500] Time 0.036 (0.039) Loss 0.8009 (1.0005) Acc@1 80.000 (74.853) Acc@5 95.000 (92.409) Test: [1950/2500] Time 0.036 (0.039) Loss 0.9170 (0.9999) Acc@1 85.000 (74.864) Acc@5 100.000 (92.414) Test: [1960/2500] Time 0.036 (0.038) Loss 3.3301 (1.0005) Acc@1 40.000 (74.834) Acc@5 60.000 (92.409) Test: [1970/2500] Time 0.036 (0.038) Loss 0.3372 (1.0024) Acc@1 95.000 (74.805) Acc@5 100.000 (92.374) Test: [1980/2500] Time 0.036 (0.038) Loss 1.9291 (1.0021) Acc@1 75.000 (74.816) Acc@5 80.000 (92.373) Test: [1990/2500] Time 0.036 (0.038) Loss 1.6760 (1.0043) Acc@1 70.000 (74.782) Acc@5 90.000 (92.343) Test: [2000/2500] Time 0.036 (0.038) Loss 0.3804 (1.0079) Acc@1 90.000 (74.730) Acc@5 95.000 (92.304) Test: [2010/2500] Time 0.036 (0.038) Loss 1.0981 (1.0049) Acc@1 80.000 (74.799) Acc@5 90.000 (92.330) Test: [2020/2500] Time 0.036 (0.038) Loss 0.8955 (1.0058) Acc@1 60.000 (74.790) Acc@5 100.000 (92.316) Test: [2030/2500] Time 0.036 (0.038) Loss 0.5169 (1.0091) Acc@1 85.000 (74.677) Acc@5 100.000 (92.277) Test: [2040/2500] Time 0.036 (0.038) Loss 0.2488 (1.0093) Acc@1 90.000 (74.672) Acc@5 100.000 (92.278) Test: [2050/2500] Time 0.036 (0.038) Loss 0.2010 (1.0113) Acc@1 90.000 (74.637) Acc@5 100.000 (92.253) Test: [2060/2500] Time 0.036 (0.038) Loss 1.3246 (1.0102) Acc@1 40.000 (74.653) Acc@5 95.000 (92.259) Test: [2070/2500] Time 0.036 (0.038) Loss 1.5121 (1.0126) Acc@1 70.000 (74.602) Acc@5 85.000 (92.240) Test: [2080/2500] Time 0.036 (0.038) Loss 0.4520 (1.0139) Acc@1 90.000 (74.589) Acc@5 95.000 (92.213) Test: [2090/2500] Time 0.037 (0.038) Loss 2.7923 (1.0155) Acc@1 30.000 (74.548) Acc@5 65.000 (92.183) Test: [2100/2500] Time 0.036 (0.038) Loss 0.9413 (1.0209) Acc@1 70.000 (74.415) Acc@5 85.000 (92.130) Test: [2110/2500] Time 0.036 (0.038) Loss 1.1209 (1.0227) Acc@1 75.000 (74.375) Acc@5 95.000 (92.108) Test: [2120/2500] Time 0.036 (0.038) Loss 1.8295 (1.0244) Acc@1 25.000 (74.331) Acc@5 95.000 (92.082) Test: [2130/2500] Time 0.036 (0.038) Loss 1.1758 (1.0248) Acc@1 75.000 (74.310) Acc@5 95.000 (92.091) Test: [2140/2500] Time 0.036 (0.038) Loss 1.4744 (1.0243) Acc@1 55.000 (74.330) Acc@5 95.000 (92.090) Test: [2150/2500] Time 0.036 (0.038) Loss 2.6390 (1.0261) Acc@1 45.000 (74.291) Acc@5 65.000 (92.069) Test: [2160/2500] Time 0.036 (0.038) Loss 0.7560 (1.0266) Acc@1 80.000 (74.271) Acc@5 90.000 (92.055) Test: [2170/2500] Time 0.036 (0.038) Loss 2.6960 (1.0273) Acc@1 60.000 (74.251) Acc@5 80.000 (92.059) Test: [2180/2500] Time 0.036 (0.038) Loss 1.4446 (1.0289) Acc@1 65.000 (74.216) Acc@5 90.000 (92.045) Test: [2190/2500] Time 0.036 (0.038) Loss 1.8486 (1.0280) Acc@1 35.000 (74.233) Acc@5 80.000 (92.056) Test: [2200/2500] Time 0.036 (0.038) Loss 0.2214 (1.0277) Acc@1 90.000 (74.230) Acc@5 100.000 (92.056) Test: [2210/2500] Time 0.036 (0.038) Loss 0.7248 (1.0290) Acc@1 80.000 (74.218) Acc@5 100.000 (92.033) Test: [2220/2500] Time 0.036 (0.038) Loss 1.1384 (1.0331) Acc@1 70.000 (74.156) Acc@5 90.000 (91.981) Test: [2230/2500] Time 0.036 (0.038) Loss 1.7130 (1.0331) Acc@1 70.000 (74.171) Acc@5 80.000 (91.979) Test: [2240/2500] Time 0.036 (0.038) Loss 0.4523 (1.0333) Acc@1 80.000 (74.139) Acc@5 100.000 (91.992) Test: [2250/2500] Time 0.036 (0.038) Loss 0.1760 (1.0358) Acc@1 95.000 (74.076) Acc@5 100.000 (91.972) Test: [2260/2500] Time 0.036 (0.038) Loss 1.4171 (1.0359) Acc@1 75.000 (74.089) Acc@5 85.000 (91.964) Test: [2270/2500] Time 0.036 (0.038) Loss 1.0877 (1.0397) Acc@1 55.000 (74.000) Acc@5 100.000 (91.933) Test: [2280/2500] Time 0.036 (0.038) Loss 0.5186 (1.0431) Acc@1 85.000 (73.915) Acc@5 100.000 (91.918) Test: [2290/2500] Time 0.036 (0.038) Loss 0.1618 (1.0413) Acc@1 100.000 (73.961) Acc@5 100.000 (91.936) Test: [2300/2500] Time 0.036 (0.038) Loss 1.3685 (1.0414) Acc@1 70.000 (73.968) Acc@5 85.000 (91.932) Test: [2310/2500] Time 0.036 (0.038) Loss 1.1738 (1.0426) Acc@1 65.000 (73.936) Acc@5 85.000 (91.928) Test: [2320/2500] Time 0.036 (0.038) Loss 1.0679 (1.0415) Acc@1 60.000 (73.953) Acc@5 95.000 (91.950) Test: [2330/2500] Time 0.036 (0.038) Loss 1.3029 (1.0432) Acc@1 70.000 (73.921) Acc@5 85.000 (91.931) Test: [2340/2500] Time 0.036 (0.038) Loss 0.2101 (1.0419) Acc@1 95.000 (73.958) Acc@5 100.000 (91.937) Test: [2350/2500] Time 0.079 (0.038) Loss 0.8369 (1.0395) Acc@1 60.000 (73.994) Acc@5 100.000 (91.963) Test: [2360/2500] Time 0.036 (0.038) Loss 0.9618 (1.0386) Acc@1 75.000 (74.003) Acc@5 90.000 (91.980) Test: [2370/2500] Time 0.036 (0.038) Loss 0.3415 (1.0377) Acc@1 90.000 (74.009) Acc@5 95.000 (91.999) Test: [2380/2500] Time 0.036 (0.038) Loss 1.0743 (1.0370) Acc@1 80.000 (74.019) Acc@5 85.000 (92.003) Test: [2390/2500] Time 0.036 (0.038) Loss 0.2286 (1.0347) Acc@1 95.000 (74.078) Acc@5 100.000 (92.022) Test: [2400/2500] Time 0.036 (0.038) Loss 2.0046 (1.0339) Acc@1 60.000 (74.100) Acc@5 80.000 (92.026) Test: [2410/2500] Time 0.036 (0.038) Loss 1.8734 (1.0380) Acc@1 55.000 (74.019) Acc@5 80.000 (91.966) Test: [2420/2500] Time 0.036 (0.038) Loss 2.5639 (1.0400) Acc@1 35.000 (73.982) Acc@5 70.000 (91.945) Test: [2430/2500] Time 0.036 (0.038) Loss 1.1588 (1.0429) Acc@1 60.000 (73.908) Acc@5 100.000 (91.909) Test: [2440/2500] Time 0.036 (0.038) Loss 0.9905 (1.0431) Acc@1 75.000 (73.898) Acc@5 95.000 (91.930) Test: [2450/2500] Time 0.036 (0.038) Loss 0.2587 (1.0450) Acc@1 95.000 (73.837) Acc@5 95.000 (91.930) Test: [2460/2500] Time 0.036 (0.038) Loss 0.2842 (1.0445) Acc@1 90.000 (73.862) Acc@5 100.000 (91.934) Test: [2470/2500] Time 0.036 (0.038) Loss 0.3358 (1.0420) Acc@1 95.000 (73.909) Acc@5 100.000 (91.951) Test: [2480/2500] Time 0.037 (0.038) Loss 0.7659 (1.0391) Acc@1 90.000 (73.986) Acc@5 90.000 (91.973) Test: [2490/2500] Time 0.036 (0.038) Loss 0.4934 (1.0366) Acc@1 85.000 (74.045) Acc@5 100.000 (91.997)
What do you mean by top 1 here?
nush12
commented
4 years ago nush12
commented
4 years ago
Hello, we tried to reproduce the ResNet50 inference, but only got 73+% top1, it is still far from SOTA(75%) and your data(76%). Any comments here?
Details: torch (1.1.0a0+a3933b8) torchvision (0.2.2.post3)
(py3-intel-chainer) [mingxiao@mlt-gpu207 examples]$ python imagenet_eval.py --data ~/data/ -a resnet50 -b 20 -e => creating model 'resnet50' => using pre-trained parameters 'imagenet' Images transformed from size 256 to [3, 224, 224] Test: [0/2500] Time 2.704 (2.704) Loss 0.7271 (0.7271) Acc@1 85.000 (85.000) Acc@5 90.000 (90.000) Test: [10/2500] Time 0.037 (0.279) Loss 0.7189 (0.5648) Acc@1 75.000 (82.727) Acc@5 95.000 (96.364) Test: [20/2500] Time 0.036 (0.164) Loss 0.9403 (0.7057) Acc@1 80.000 (80.714) Acc@5 90.000 (94.762) Test: [30/2500] Time 0.036 (0.123) Loss 0.5037 (0.6162) Acc@1 90.000 (83.871) Acc@5 95.000 (95.645) Test: [40/2500] Time 0.036 (0.102) Loss 0.7368 (0.5309) Acc@1 85.000 (85.976) Acc@5 85.000 (95.854) Test: [50/2500] Time 0.036 (0.089) Loss 0.6895 (0.4964) Acc@1 80.000 (86.961) Acc@5 90.000 (96.078) Test: [60/2500] Time 0.036 (0.080) Loss 0.7518 (0.5163) Acc@1 90.000 (86.967) Acc@5 95.000 (95.820) Test: [70/2500] Time 0.036 (0.074) Loss 0.2763 (0.5536) Acc@1 90.000 (85.845) Acc@5 100.000 (95.986) Test: [80/2500] Time 0.037 (0.069) Loss 2.0607 (0.6061) Acc@1 40.000 (85.185) Acc@5 90.000 (95.370) Test: [90/2500] Time 0.036 (0.066) Loss 1.9135 (0.7201) Acc@1 40.000 (82.253) Acc@5 95.000 (94.176) Test: [100/2500] Time 0.036 (0.063) Loss 0.7474 (0.7224) Acc@1 50.000 (81.634) Acc@5 100.000 (94.505) Test: [110/2500] Time 0.036 (0.060) Loss 1.3561 (0.7373) Acc@1 65.000 (80.946) Acc@5 95.000 (94.459) Test: [120/2500] Time 0.037 (0.059) Loss 0.4951 (0.7848) Acc@1 90.000 (79.959) Acc@5 100.000 (94.008) Test: [130/2500] Time 0.037 (0.057) Loss 0.5267 (0.7856) Acc@1 80.000 (79.924) Acc@5 100.000 (93.893) Test: [140/2500] Time 0.036 (0.055) Loss 0.8788 (0.8200) Acc@1 75.000 (79.291) Acc@5 95.000 (93.617) Test: [150/2500] Time 0.036 (0.054) Loss 2.6851 (0.8557) Acc@1 35.000 (78.543) Acc@5 85.000 (93.245) Test: [160/2500] Time 0.036 (0.053) Loss 2.0516 (0.8964) Acc@1 50.000 (77.422) Acc@5 90.000 (93.043) Test: [170/2500] Time 0.036 (0.052) Loss 0.9259 (0.9198) Acc@1 55.000 (76.579) Acc@5 100.000 (92.865) Test: [180/2500] Time 0.036 (0.051) Loss 0.6522 (0.9092) Acc@1 75.000 (76.878) Acc@5 100.000 (92.956) Test: [190/2500] Time 0.036 (0.050) Loss 0.2513 (0.9095) Acc@1 85.000 (76.230) Acc@5 100.000 (93.141) Test: [200/2500] Time 0.037 (0.050) Loss 0.2398 (0.9087) Acc@1 95.000 (76.567) Acc@5 100.000 (93.159) Test: [210/2500] Time 0.037 (0.049) Loss 0.3800 (0.8916) Acc@1 90.000 (77.038) Acc@5 95.000 (93.294) Test: [220/2500] Time 0.036 (0.049) Loss 0.6269 (0.8736) Acc@1 90.000 (77.511) Acc@5 95.000 (93.462) Test: [230/2500] Time 0.036 (0.048) Loss 0.0200 (0.8483) Acc@1 100.000 (78.225) Acc@5 100.000 (93.615) Test: [240/2500] Time 0.036 (0.048) Loss 0.0885 (0.8253) Acc@1 95.000 (78.900) Acc@5 100.000 (93.797) Test: [250/2500] Time 0.037 (0.047) Loss 0.0069 (0.8111) Acc@1 100.000 (79.283) Acc@5 100.000 (93.944) Test: [260/2500] Time 0.036 (0.047) Loss 0.3296 (0.8053) Acc@1 95.000 (79.368) Acc@5 95.000 (93.946) Test: [270/2500] Time 0.036 (0.046) Loss 1.4653 (0.8146) Acc@1 60.000 (79.299) Acc@5 90.000 (93.875) Test: [280/2500] Time 0.036 (0.046) Loss 1.3985 (0.8086) Acc@1 65.000 (79.395) Acc@5 80.000 (93.932) Test: [290/2500] Time 0.037 (0.046) Loss 0.3668 (0.8072) Acc@1 90.000 (79.519) Acc@5 100.000 (93.883) Test: [300/2500] Time 0.037 (0.045) Loss 0.7728 (0.8094) Acc@1 80.000 (79.535) Acc@5 95.000 (93.920) Test: [310/2500] Time 0.037 (0.045) Loss 1.3635 (0.8137) Acc@1 65.000 (79.357) Acc@5 95.000 (93.955) Test: [320/2500] Time 0.036 (0.045) Loss 0.8331 (0.8190) Acc@1 85.000 (79.346) Acc@5 95.000 (93.941) Test: [330/2500] Time 0.036 (0.045) Loss 0.4665 (0.8053) Acc@1 90.000 (79.653) Acc@5 95.000 (94.033) Test: [340/2500] Time 0.037 (0.044) Loss 0.0864 (0.7940) Acc@1 95.000 (79.941) Acc@5 100.000 (94.120) Test: [350/2500] Time 0.036 (0.044) Loss 1.1962 (0.7807) Acc@1 75.000 (80.271) Acc@5 95.000 (94.231) Test: [360/2500] Time 0.036 (0.044) Loss 0.2139 (0.7706) Acc@1 95.000 (80.512) Acc@5 95.000 (94.294) Test: [370/2500] Time 0.037 (0.044) Loss 0.1379 (0.7592) Acc@1 95.000 (80.809) Acc@5 100.000 (94.394) Test: [380/2500] Time 0.036 (0.043) Loss 1.0197 (0.7602) Acc@1 75.000 (80.866) Acc@5 90.000 (94.370) Test: [390/2500] Time 0.037 (0.043) Loss 0.1889 (0.7601) Acc@1 95.000 (80.895) Acc@5 100.000 (94.412) Test: [400/2500] Time 0.036 (0.043) Loss 0.2492 (0.7598) Acc@1 90.000 (80.885) Acc@5 100.000 (94.439) Test: [410/2500] Time 0.037 (0.043) Loss 0.2367 (0.7663) Acc@1 95.000 (80.779) Acc@5 100.000 (94.367) Test: [420/2500] Time 0.036 (0.043) Loss 1.1600 (0.7876) Acc@1 60.000 (80.036) Acc@5 85.000 (94.252) Test: [430/2500] Time 0.036 (0.043) Loss 1.1108 (0.7932) Acc@1 75.000 (79.988) Acc@5 90.000 (94.223) Test: [440/2500] Time 0.037 (0.043) Loss 0.5544 (0.7979) Acc@1 80.000 (79.796) Acc@5 100.000 (94.263) Test: [450/2500] Time 0.036 (0.042) Loss 1.0699 (0.7975) Acc@1 55.000 (79.723) Acc@5 95.000 (94.290) Test: [460/2500] Time 0.037 (0.042) Loss 1.0099 (0.7941) Acc@1 80.000 (79.783) Acc@5 90.000 (94.328) Test: [470/2500] Time 0.036 (0.042) Loss 0.6814 (0.8014) Acc@1 80.000 (79.544) Acc@5 100.000 (94.299) Test: [480/2500] Time 0.036 (0.042) Loss 0.5434 (0.8016) Acc@1 80.000 (79.511) Acc@5 100.000 (94.335) Test: [490/2500] Time 0.037 (0.042) Loss 0.7770 (0.8022) Acc@1 75.000 (79.430) Acc@5 95.000 (94.358) Test: [500/2500] Time 0.036 (0.042) Loss 1.1530 (0.8020) Acc@1 65.000 (79.401) Acc@5 100.000 (94.431) Test: [510/2500] Time 0.036 (0.042) Loss 0.9519 (0.8027) Acc@1 65.000 (79.247) Acc@5 95.000 (94.472) Test: [520/2500] Time 0.036 (0.042) Loss 0.2959 (0.8030) Acc@1 95.000 (79.165) Acc@5 100.000 (94.491) Test: [530/2500] Time 0.036 (0.042) Loss 0.8767 (0.8013) Acc@1 75.000 (79.134) Acc@5 100.000 (94.557) Test: [540/2500] Time 0.105 (0.042) Loss 0.3152 (0.8015) Acc@1 95.000 (79.168) Acc@5 100.000 (94.538) Test: [550/2500] Time 0.036 (0.042) Loss 1.6377 (0.7974) Acc@1 65.000 (79.274) Acc@5 85.000 (94.583) Test: [560/2500] Time 0.036 (0.042) Loss 0.4664 (0.7982) Acc@1 85.000 (79.269) Acc@5 95.000 (94.563) Test: [570/2500] Time 0.037 (0.042) Loss 0.6717 (0.8058) Acc@1 85.000 (79.177) Acc@5 95.000 (94.483) Test: [580/2500] Time 0.036 (0.042) Loss 0.4850 (0.8037) Acc@1 90.000 (79.157) Acc@5 100.000 (94.527) Test: [590/2500] Time 0.036 (0.042) Loss 0.7717 (0.8043) Acc@1 85.000 (79.120) Acc@5 95.000 (94.552) Test: [600/2500] Time 0.037 (0.042) Loss 2.1893 (0.8087) Acc@1 25.000 (78.993) Acc@5 100.000 (94.567) Test: [610/2500] Time 0.036 (0.041) Loss 0.0302 (0.8118) Acc@1 100.000 (78.756) Acc@5 100.000 (94.615) Test: [620/2500] Time 0.037 (0.041) Loss 1.2387 (0.8152) Acc@1 50.000 (78.647) Acc@5 95.000 (94.589) Test: [630/2500] Time 0.036 (0.041) Loss 0.5377 (0.8129) Acc@1 85.000 (78.558) Acc@5 95.000 (94.651) Test: [640/2500] Time 0.037 (0.041) Loss 0.4386 (0.8061) Acc@1 80.000 (78.736) Acc@5 100.000 (94.696) Test: [650/2500] Time 0.037 (0.041) Loss 0.5441 (0.8006) Acc@1 80.000 (78.902) Acc@5 100.000 (94.739) Test: [660/2500] Time 0.037 (0.041) Loss 0.7073 (0.7952) Acc@1 75.000 (79.032) Acc@5 100.000 (94.796) Test: [670/2500] Time 0.036 (0.041) Loss 0.3843 (0.7975) Acc@1 90.000 (78.815) Acc@5 95.000 (94.806) Test: [680/2500] Time 0.036 (0.041) Loss 2.9731 (0.8027) Acc@1 55.000 (78.744) Acc@5 90.000 (94.809) Test: [690/2500] Time 0.036 (0.041) Loss 0.7158 (0.8016) Acc@1 85.000 (78.828) Acc@5 95.000 (94.797) Test: [700/2500] Time 0.036 (0.041) Loss 0.8419 (0.8031) Acc@1 70.000 (78.738) Acc@5 95.000 (94.807) Test: [710/2500] Time 0.036 (0.041) Loss 0.0556 (0.8060) Acc@1 100.000 (78.579) Acc@5 100.000 (94.824) Test: [720/2500] Time 0.036 (0.041) Loss 0.8053 (0.8107) Acc@1 80.000 (78.530) Acc@5 100.000 (94.785) Test: [730/2500] Time 0.036 (0.041) Loss 0.7175 (0.8074) Acc@1 90.000 (78.618) Acc@5 100.000 (94.815) Test: [740/2500] Time 0.036 (0.041) Loss 0.6575 (0.8006) Acc@1 95.000 (78.806) Acc@5 95.000 (94.872) Test: [750/2500] Time 0.036 (0.041) Loss 0.2692 (0.8036) Acc@1 95.000 (78.815) Acc@5 95.000 (94.854) Test: [760/2500] Time 0.036 (0.040) Loss 0.9070 (0.8051) Acc@1 70.000 (78.739) Acc@5 100.000 (94.862) Test: [770/2500] Time 0.036 (0.040) Loss 0.5024 (0.8015) Acc@1 95.000 (78.800) Acc@5 95.000 (94.890) Test: [780/2500] Time 0.036 (0.040) Loss 1.1427 (0.8049) Acc@1 65.000 (78.713) Acc@5 95.000 (94.898) Test: [790/2500] Time 0.036 (0.040) Loss 0.7711 (0.8105) Acc@1 80.000 (78.603) Acc@5 95.000 (94.848) Test: [800/2500] Time 0.036 (0.040) Loss 0.2140 (0.8079) Acc@1 90.000 (78.658) Acc@5 100.000 (94.881) Test: [810/2500] Time 0.036 (0.040) Loss 0.1252 (0.8029) Acc@1 95.000 (78.798) Acc@5 100.000 (94.920) Test: [820/2500] Time 0.036 (0.040) Loss 0.4773 (0.7979) Acc@1 90.000 (78.959) Acc@5 95.000 (94.945) Test: [830/2500] Time 0.036 (0.040) Loss 0.0433 (0.7976) Acc@1 100.000 (78.995) Acc@5 100.000 (94.958) Test: [840/2500] Time 0.036 (0.040) Loss 0.6456 (0.7937) Acc@1 85.000 (79.132) Acc@5 95.000 (94.982) Test: [850/2500] Time 0.036 (0.040) Loss 0.3178 (0.7914) Acc@1 95.000 (79.236) Acc@5 95.000 (95.000) Test: [860/2500] Time 0.036 (0.040) Loss 0.7505 (0.7906) Acc@1 80.000 (79.228) Acc@5 90.000 (95.000) Test: [870/2500] Time 0.036 (0.040) Loss 0.5531 (0.7903) Acc@1 75.000 (79.214) Acc@5 100.000 (95.023) Test: [880/2500] Time 0.036 (0.040) Loss 0.3595 (0.7870) Acc@1 95.000 (79.257) Acc@5 100.000 (95.051) Test: [890/2500] Time 0.036 (0.040) Loss 1.7804 (0.7870) Acc@1 55.000 (79.237) Acc@5 95.000 (95.056) Test: [900/2500] Time 0.036 (0.040) Loss 0.8308 (0.7935) Acc@1 80.000 (79.057) Acc@5 90.000 (95.067) Test: [910/2500] Time 0.036 (0.040) Loss 0.0886 (0.7898) Acc@1 95.000 (79.166) Acc@5 100.000 (95.082) Test: [920/2500] Time 0.036 (0.040) Loss 1.3319 (0.7896) Acc@1 70.000 (79.224) Acc@5 95.000 (95.060) Test: [930/2500] Time 0.036 (0.040) Loss 0.0865 (0.7928) Acc@1 100.000 (79.216) Acc@5 100.000 (95.038) Test: [940/2500] Time 0.036 (0.040) Loss 0.0196 (0.7941) Acc@1 100.000 (79.203) Acc@5 100.000 (95.011) Test: [950/2500] Time 0.036 (0.040) Loss 1.4047 (0.7955) Acc@1 65.000 (79.138) Acc@5 90.000 (94.979) Test: [960/2500] Time 0.036 (0.040) Loss 0.7717 (0.7998) Acc@1 85.000 (79.022) Acc@5 95.000 (94.927) Test: [970/2500] Time 0.036 (0.040) Loss 0.2456 (0.7992) Acc@1 95.000 (79.001) Acc@5 95.000 (94.938) Test: [980/2500] Time 0.036 (0.040) Loss 0.3763 (0.8003) Acc@1 95.000 (79.006) Acc@5 95.000 (94.903) Test: [990/2500] Time 0.036 (0.039) Loss 0.3058 (0.8009) Acc@1 90.000 (78.991) Acc@5 95.000 (94.879) Test: [1000/2500] Time 0.036 (0.039) Loss 1.5105 (0.8021) Acc@1 55.000 (78.996) Acc@5 95.000 (94.880) Test: [1010/2500] Time 0.036 (0.039) Loss 0.1445 (0.8050) Acc@1 100.000 (78.937) Acc@5 100.000 (94.832) Test: [1020/2500] Time 0.036 (0.039) Loss 1.2508 (0.8037) Acc@1 70.000 (78.962) Acc@5 95.000 (94.853) Test: [1030/2500] Time 0.036 (0.039) Loss 1.7219 (0.8058) Acc@1 65.000 (78.923) Acc@5 80.000 (94.830) Test: [1040/2500] Time 0.036 (0.039) Loss 0.2517 (0.8198) Acc@1 100.000 (78.612) Acc@5 100.000 (94.673) Test: [1050/2500] Time 0.036 (0.039) Loss 0.1276 (0.8206) Acc@1 100.000 (78.578) Acc@5 100.000 (94.676) Test: [1060/2500] Time 0.036 (0.039) Loss 0.9815 (0.8214) Acc@1 70.000 (78.530) Acc@5 100.000 (94.694) Test: [1070/2500] Time 0.036 (0.039) Loss 1.5380 (0.8224) Acc@1 65.000 (78.492) Acc@5 90.000 (94.664) Test: [1080/2500] Time 0.036 (0.039) Loss 0.5527 (0.8213) Acc@1 80.000 (78.511) Acc@5 100.000 (94.672) Test: [1090/2500] Time 0.036 (0.039) Loss 1.7449 (0.8259) Acc@1 65.000 (78.368) Acc@5 85.000 (94.615) Test: [1100/2500] Time 0.036 (0.039) Loss 0.8507 (0.8296) Acc@1 75.000 (78.333) Acc@5 95.000 (94.569) Test: [1110/2500] Time 0.036 (0.039) Loss 0.6621 (0.8307) Acc@1 85.000 (78.281) Acc@5 100.000 (94.568) Test: [1120/2500] Time 0.036 (0.039) Loss 0.5863 (0.8357) Acc@1 80.000 (78.158) Acc@5 90.000 (94.514) Test: [1130/2500] Time 0.036 (0.039) Loss 1.4977 (0.8335) Acc@1 70.000 (78.223) Acc@5 90.000 (94.527) Test: [1140/2500] Time 0.036 (0.039) Loss 2.4943 (0.8396) Acc@1 50.000 (78.124) Acc@5 75.000 (94.452) Test: [1150/2500] Time 0.036 (0.039) Loss 1.2823 (0.8440) Acc@1 70.000 (78.076) Acc@5 90.000 (94.387) Test: [1160/2500] Time 0.036 (0.039) Loss 1.7010 (0.8497) Acc@1 55.000 (77.929) Acc@5 85.000 (94.337) Test: [1170/2500] Time 0.036 (0.039) Loss 0.9460 (0.8530) Acc@1 70.000 (77.852) Acc@5 90.000 (94.270) Test: [1180/2500] Time 0.036 (0.039) Loss 0.1823 (0.8582) Acc@1 95.000 (77.786) Acc@5 100.000 (94.191) Test: [1190/2500] Time 0.036 (0.039) Loss 0.0599 (0.8573) Acc@1 95.000 (77.813) Acc@5 100.000 (94.190) Test: [1200/2500] Time 0.036 (0.039) Loss 0.6744 (0.8641) Acc@1 80.000 (77.664) Acc@5 90.000 (94.105) Test: [1210/2500] Time 0.036 (0.039) Loss 0.7390 (0.8671) Acc@1 65.000 (77.519) Acc@5 100.000 (94.075) Test: [1220/2500] Time 0.036 (0.039) Loss 1.6444 (0.8700) Acc@1 50.000 (77.416) Acc@5 100.000 (94.054) Test: [1230/2500] Time 0.036 (0.039) Loss 1.2105 (0.8737) Acc@1 70.000 (77.360) Acc@5 95.000 (94.025) Test: [1240/2500] Time 0.036 (0.039) Loss 0.5262 (0.8801) Acc@1 85.000 (77.236) Acc@5 90.000 (93.924) Test: [1250/2500] Time 0.036 (0.039) Loss 0.3565 (0.8854) Acc@1 85.000 (77.142) Acc@5 100.000 (93.861) Test: [1260/2500] Time 0.036 (0.039) Loss 1.7938 (0.8913) Acc@1 45.000 (77.006) Acc@5 80.000 (93.775) Test: [1270/2500] Time 0.036 (0.039) Loss 0.9489 (0.8963) Acc@1 80.000 (76.904) Acc@5 95.000 (93.729) Test: [1280/2500] Time 0.036 (0.039) Loss 0.2843 (0.8965) Acc@1 100.000 (76.874) Acc@5 100.000 (93.716) Test: [1290/2500] Time 0.036 (0.039) Loss 2.0598 (0.8986) Acc@1 45.000 (76.824) Acc@5 85.000 (93.718) Test: [1300/2500] Time 0.036 (0.039) Loss 0.5175 (0.9018) Acc@1 90.000 (76.791) Acc@5 100.000 (93.689) Test: [1310/2500] Time 0.077 (0.039) Loss 1.3497 (0.9046) Acc@1 40.000 (76.709) Acc@5 95.000 (93.677) Test: [1320/2500] Time 0.036 (0.039) Loss 0.9780 (0.9085) Acc@1 65.000 (76.578) Acc@5 85.000 (93.656) Test: [1330/2500] Time 0.036 (0.039) Loss 0.9246 (0.9132) Acc@1 70.000 (76.476) Acc@5 90.000 (93.621) Test: [1340/2500] Time 0.036 (0.039) Loss 2.1366 (0.9145) Acc@1 55.000 (76.454) Acc@5 85.000 (93.594) Test: [1350/2500] Time 0.036 (0.039) Loss 0.3810 (0.9145) Acc@1 90.000 (76.429) Acc@5 95.000 (93.601) Test: [1360/2500] Time 0.036 (0.039) Loss 0.5766 (0.9179) Acc@1 90.000 (76.356) Acc@5 100.000 (93.564) Test: [1370/2500] Time 0.036 (0.039) Loss 0.2055 (0.9168) Acc@1 95.000 (76.371) Acc@5 100.000 (93.567) Test: [1380/2500] Time 0.070 (0.039) Loss 0.2993 (0.9179) Acc@1 90.000 (76.336) Acc@5 95.000 (93.559) Test: [1390/2500] Time 0.036 (0.039) Loss 1.2469 (0.9162) Acc@1 70.000 (76.387) Acc@5 95.000 (93.562) Test: [1400/2500] Time 0.036 (0.039) Loss 0.2995 (0.9225) Acc@1 95.000 (76.299) Acc@5 95.000 (93.487) Test: [1410/2500] Time 0.036 (0.039) Loss 0.4661 (0.9195) Acc@1 85.000 (76.364) Acc@5 95.000 (93.505) Test: [1420/2500] Time 0.036 (0.039) Loss 0.3127 (0.9195) Acc@1 95.000 (76.383) Acc@5 100.000 (93.487) Test: [1430/2500] Time 0.036 (0.039) Loss 1.5971 (0.9186) Acc@1 60.000 (76.433) Acc@5 80.000 (93.484) Test: [1440/2500] Time 0.036 (0.039) Loss 0.0613 (0.9173) Acc@1 100.000 (76.492) Acc@5 100.000 (93.487) Test: [1450/2500] Time 0.036 (0.039) Loss 0.5770 (0.9166) Acc@1 80.000 (76.465) Acc@5 100.000 (93.501) Test: [1460/2500] Time 0.036 (0.039) Loss 2.6388 (0.9189) Acc@1 45.000 (76.451) Acc@5 55.000 (93.474) Test: [1470/2500] Time 0.036 (0.039) Loss 1.0036 (0.9248) Acc@1 85.000 (76.360) Acc@5 90.000 (93.402) Test: [1480/2500] Time 0.036 (0.039) Loss 0.8197 (0.9289) Acc@1 80.000 (76.283) Acc@5 95.000 (93.352) Test: [1490/2500] Time 0.036 (0.039) Loss 1.8379 (0.9277) Acc@1 60.000 (76.328) Acc@5 85.000 (93.357) Test: [1500/2500] Time 0.036 (0.039) Loss 2.8159 (0.9339) Acc@1 40.000 (76.219) Acc@5 75.000 (93.294) Test: [1510/2500] Time 0.036 (0.039) Loss 0.0468 (0.9372) Acc@1 100.000 (76.168) Acc@5 100.000 (93.250) Test: [1520/2500] Time 0.036 (0.039) Loss 1.1193 (0.9350) Acc@1 75.000 (76.239) Acc@5 95.000 (93.261) Test: [1530/2500] Time 0.036 (0.039) Loss 0.3309 (0.9345) Acc@1 95.000 (76.267) Acc@5 95.000 (93.246) Test: [1540/2500] Time 0.036 (0.039) Loss 1.8861 (0.9341) Acc@1 75.000 (76.298) Acc@5 80.000 (93.228) Test: [1550/2500] Time 0.036 (0.039) Loss 1.8906 (0.9383) Acc@1 35.000 (76.167) Acc@5 85.000 (93.182) Test: [1560/2500] Time 0.036 (0.039) Loss 1.3631 (0.9441) Acc@1 60.000 (76.019) Acc@5 90.000 (93.107) Test: [1570/2500] Time 0.036 (0.039) Loss 0.3538 (0.9437) Acc@1 85.000 (76.047) Acc@5 100.000 (93.090) Test: [1580/2500] Time 0.036 (0.039) Loss 1.6606 (0.9441) Acc@1 60.000 (76.031) Acc@5 80.000 (93.080) Test: [1590/2500] Time 0.036 (0.039) Loss 0.7779 (0.9503) Acc@1 80.000 (75.889) Acc@5 100.000 (93.011) Test: [1600/2500] Time 0.036 (0.039) Loss 0.2365 (0.9523) Acc@1 95.000 (75.731) Acc@5 100.000 (93.017) Test: [1610/2500] Time 0.036 (0.039) Loss 0.9663 (0.9524) Acc@1 70.000 (75.739) Acc@5 95.000 (93.014) Test: [1620/2500] Time 0.036 (0.039) Loss 1.4808 (0.9528) Acc@1 65.000 (75.762) Acc@5 90.000 (93.001) Test: [1630/2500] Time 0.036 (0.039) Loss 1.9637 (0.9574) Acc@1 60.000 (75.696) Acc@5 75.000 (92.934) Test: [1640/2500] Time 0.036 (0.039) Loss 2.0634 (0.9611) Acc@1 55.000 (75.637) Acc@5 90.000 (92.895) Test: [1650/2500] Time 0.036 (0.039) Loss 1.1165 (0.9642) Acc@1 75.000 (75.533) Acc@5 90.000 (92.856) Test: [1660/2500] Time 0.037 (0.039) Loss 2.1994 (0.9665) Acc@1 30.000 (75.458) Acc@5 80.000 (92.839) Test: [1670/2500] Time 0.037 (0.039) Loss 0.4767 (0.9670) Acc@1 85.000 (75.416) Acc@5 100.000 (92.849) Test: [1680/2500] Time 0.036 (0.039) Loss 0.3436 (0.9643) Acc@1 90.000 (75.467) Acc@5 95.000 (92.879) Test: [1690/2500] Time 0.036 (0.039) Loss 2.8868 (0.9682) Acc@1 45.000 (75.387) Acc@5 75.000 (92.850) Test: [1700/2500] Time 0.036 (0.039) Loss 0.9127 (0.9718) Acc@1 75.000 (75.373) Acc@5 95.000 (92.798) Test: [1710/2500] Time 0.036 (0.039) Loss 1.3394 (0.9741) Acc@1 70.000 (75.324) Acc@5 85.000 (92.779) Test: [1720/2500] Time 0.036 (0.039) Loss 0.0846 (0.9728) Acc@1 100.000 (75.369) Acc@5 100.000 (92.780) Test: [1730/2500] Time 0.036 (0.039) Loss 3.2437 (0.9778) Acc@1 30.000 (75.257) Acc@5 65.000 (92.733) Test: [1740/2500] Time 0.036 (0.039) Loss 2.0856 (0.9784) Acc@1 80.000 (75.253) Acc@5 85.000 (92.720) Test: [1750/2500] Time 0.036 (0.039) Loss 0.9843 (0.9790) Acc@1 75.000 (75.237) Acc@5 90.000 (92.704) Test: [1760/2500] Time 0.036 (0.039) Loss 0.6655 (0.9791) Acc@1 80.000 (75.233) Acc@5 95.000 (92.712) Test: [1770/2500] Time 0.036 (0.039) Loss 1.2545 (0.9809) Acc@1 65.000 (75.189) Acc@5 90.000 (92.699) Test: [1780/2500] Time 0.036 (0.039) Loss 1.7366 (0.9827) Acc@1 70.000 (75.143) Acc@5 85.000 (92.681) Test: [1790/2500] Time 0.036 (0.039) Loss 0.8464 (0.9822) Acc@1 95.000 (75.179) Acc@5 95.000 (92.683) Test: [1800/2500] Time 0.036 (0.039) Loss 0.7508 (0.9816) Acc@1 75.000 (75.183) Acc@5 100.000 (92.685) Test: [1810/2500] Time 0.036 (0.039) Loss 0.5370 (0.9809) Acc@1 80.000 (75.215) Acc@5 100.000 (92.681) Test: [1820/2500] Time 0.036 (0.039) Loss 3.9797 (0.9830) Acc@1 35.000 (75.181) Acc@5 60.000 (92.652) Test: [1830/2500] Time 0.036 (0.039) Loss 0.7985 (0.9887) Acc@1 80.000 (75.090) Acc@5 90.000 (92.567) Test: [1840/2500] Time 0.036 (0.039) Loss 1.1496 (0.9899) Acc@1 80.000 (75.062) Acc@5 85.000 (92.539) Test: [1850/2500] Time 0.036 (0.039) Loss 2.3159 (0.9894) Acc@1 55.000 (75.092) Acc@5 70.000 (92.545) Test: [1860/2500] Time 0.036 (0.039) Loss 1.4142 (0.9915) Acc@1 50.000 (75.032) Acc@5 90.000 (92.517) Test: [1870/2500] Time 0.036 (0.039) Loss 1.5630 (0.9932) Acc@1 50.000 (74.992) Acc@5 90.000 (92.496) Test: [1880/2500] Time 0.036 (0.039) Loss 1.0481 (0.9947) Acc@1 80.000 (74.942) Acc@5 85.000 (92.485) Test: [1890/2500] Time 0.041 (0.039) Loss 0.4099 (0.9943) Acc@1 90.000 (74.960) Acc@5 100.000 (92.491) Test: [1900/2500] Time 0.036 (0.039) Loss 1.0773 (0.9940) Acc@1 75.000 (74.961) Acc@5 95.000 (92.501) Test: [1910/2500] Time 0.036 (0.039) Loss 2.7580 (0.9963) Acc@1 25.000 (74.927) Acc@5 80.000 (92.475) Test: [1920/2500] Time 0.036 (0.039) Loss 0.9038 (0.9986) Acc@1 80.000 (74.901) Acc@5 90.000 (92.431) Test: [1930/2500] Time 0.036 (0.039) Loss 1.7654 (0.9984) Acc@1 45.000 (74.904) Acc@5 80.000 (92.431) Test: [1940/2500] Time 0.036 (0.039) Loss 0.8009 (1.0005) Acc@1 80.000 (74.853) Acc@5 95.000 (92.409) Test: [1950/2500] Time 0.036 (0.039) Loss 0.9170 (0.9999) Acc@1 85.000 (74.864) Acc@5 100.000 (92.414) Test: [1960/2500] Time 0.036 (0.038) Loss 3.3301 (1.0005) Acc@1 40.000 (74.834) Acc@5 60.000 (92.409) Test: [1970/2500] Time 0.036 (0.038) Loss 0.3372 (1.0024) Acc@1 95.000 (74.805) Acc@5 100.000 (92.374) Test: [1980/2500] Time 0.036 (0.038) Loss 1.9291 (1.0021) Acc@1 75.000 (74.816) Acc@5 80.000 (92.373) Test: [1990/2500] Time 0.036 (0.038) Loss 1.6760 (1.0043) Acc@1 70.000 (74.782) Acc@5 90.000 (92.343) Test: [2000/2500] Time 0.036 (0.038) Loss 0.3804 (1.0079) Acc@1 90.000 (74.730) Acc@5 95.000 (92.304) Test: [2010/2500] Time 0.036 (0.038) Loss 1.0981 (1.0049) Acc@1 80.000 (74.799) Acc@5 90.000 (92.330) Test: [2020/2500] Time 0.036 (0.038) Loss 0.8955 (1.0058) Acc@1 60.000 (74.790) Acc@5 100.000 (92.316) Test: [2030/2500] Time 0.036 (0.038) Loss 0.5169 (1.0091) Acc@1 85.000 (74.677) Acc@5 100.000 (92.277) Test: [2040/2500] Time 0.036 (0.038) Loss 0.2488 (1.0093) Acc@1 90.000 (74.672) Acc@5 100.000 (92.278) Test: [2050/2500] Time 0.036 (0.038) Loss 0.2010 (1.0113) Acc@1 90.000 (74.637) Acc@5 100.000 (92.253) Test: [2060/2500] Time 0.036 (0.038) Loss 1.3246 (1.0102) Acc@1 40.000 (74.653) Acc@5 95.000 (92.259) Test: [2070/2500] Time 0.036 (0.038) Loss 1.5121 (1.0126) Acc@1 70.000 (74.602) Acc@5 85.000 (92.240) Test: [2080/2500] Time 0.036 (0.038) Loss 0.4520 (1.0139) Acc@1 90.000 (74.589) Acc@5 95.000 (92.213) Test: [2090/2500] Time 0.037 (0.038) Loss 2.7923 (1.0155) Acc@1 30.000 (74.548) Acc@5 65.000 (92.183) Test: [2100/2500] Time 0.036 (0.038) Loss 0.9413 (1.0209) Acc@1 70.000 (74.415) Acc@5 85.000 (92.130) Test: [2110/2500] Time 0.036 (0.038) Loss 1.1209 (1.0227) Acc@1 75.000 (74.375) Acc@5 95.000 (92.108) Test: [2120/2500] Time 0.036 (0.038) Loss 1.8295 (1.0244) Acc@1 25.000 (74.331) Acc@5 95.000 (92.082) Test: [2130/2500] Time 0.036 (0.038) Loss 1.1758 (1.0248) Acc@1 75.000 (74.310) Acc@5 95.000 (92.091) Test: [2140/2500] Time 0.036 (0.038) Loss 1.4744 (1.0243) Acc@1 55.000 (74.330) Acc@5 95.000 (92.090) Test: [2150/2500] Time 0.036 (0.038) Loss 2.6390 (1.0261) Acc@1 45.000 (74.291) Acc@5 65.000 (92.069) Test: [2160/2500] Time 0.036 (0.038) Loss 0.7560 (1.0266) Acc@1 80.000 (74.271) Acc@5 90.000 (92.055) Test: [2170/2500] Time 0.036 (0.038) Loss 2.6960 (1.0273) Acc@1 60.000 (74.251) Acc@5 80.000 (92.059) Test: [2180/2500] Time 0.036 (0.038) Loss 1.4446 (1.0289) Acc@1 65.000 (74.216) Acc@5 90.000 (92.045) Test: [2190/2500] Time 0.036 (0.038) Loss 1.8486 (1.0280) Acc@1 35.000 (74.233) Acc@5 80.000 (92.056) Test: [2200/2500] Time 0.036 (0.038) Loss 0.2214 (1.0277) Acc@1 90.000 (74.230) Acc@5 100.000 (92.056) Test: [2210/2500] Time 0.036 (0.038) Loss 0.7248 (1.0290) Acc@1 80.000 (74.218) Acc@5 100.000 (92.033) Test: [2220/2500] Time 0.036 (0.038) Loss 1.1384 (1.0331) Acc@1 70.000 (74.156) Acc@5 90.000 (91.981) Test: [2230/2500] Time 0.036 (0.038) Loss 1.7130 (1.0331) Acc@1 70.000 (74.171) Acc@5 80.000 (91.979) Test: [2240/2500] Time 0.036 (0.038) Loss 0.4523 (1.0333) Acc@1 80.000 (74.139) Acc@5 100.000 (91.992) Test: [2250/2500] Time 0.036 (0.038) Loss 0.1760 (1.0358) Acc@1 95.000 (74.076) Acc@5 100.000 (91.972) Test: [2260/2500] Time 0.036 (0.038) Loss 1.4171 (1.0359) Acc@1 75.000 (74.089) Acc@5 85.000 (91.964) Test: [2270/2500] Time 0.036 (0.038) Loss 1.0877 (1.0397) Acc@1 55.000 (74.000) Acc@5 100.000 (91.933) Test: [2280/2500] Time 0.036 (0.038) Loss 0.5186 (1.0431) Acc@1 85.000 (73.915) Acc@5 100.000 (91.918) Test: [2290/2500] Time 0.036 (0.038) Loss 0.1618 (1.0413) Acc@1 100.000 (73.961) Acc@5 100.000 (91.936) Test: [2300/2500] Time 0.036 (0.038) Loss 1.3685 (1.0414) Acc@1 70.000 (73.968) Acc@5 85.000 (91.932) Test: [2310/2500] Time 0.036 (0.038) Loss 1.1738 (1.0426) Acc@1 65.000 (73.936) Acc@5 85.000 (91.928) Test: [2320/2500] Time 0.036 (0.038) Loss 1.0679 (1.0415) Acc@1 60.000 (73.953) Acc@5 95.000 (91.950) Test: [2330/2500] Time 0.036 (0.038) Loss 1.3029 (1.0432) Acc@1 70.000 (73.921) Acc@5 85.000 (91.931) Test: [2340/2500] Time 0.036 (0.038) Loss 0.2101 (1.0419) Acc@1 95.000 (73.958) Acc@5 100.000 (91.937) Test: [2350/2500] Time 0.079 (0.038) Loss 0.8369 (1.0395) Acc@1 60.000 (73.994) Acc@5 100.000 (91.963) Test: [2360/2500] Time 0.036 (0.038) Loss 0.9618 (1.0386) Acc@1 75.000 (74.003) Acc@5 90.000 (91.980) Test: [2370/2500] Time 0.036 (0.038) Loss 0.3415 (1.0377) Acc@1 90.000 (74.009) Acc@5 95.000 (91.999) Test: [2380/2500] Time 0.036 (0.038) Loss 1.0743 (1.0370) Acc@1 80.000 (74.019) Acc@5 85.000 (92.003) Test: [2390/2500] Time 0.036 (0.038) Loss 0.2286 (1.0347) Acc@1 95.000 (74.078) Acc@5 100.000 (92.022) Test: [2400/2500] Time 0.036 (0.038) Loss 2.0046 (1.0339) Acc@1 60.000 (74.100) Acc@5 80.000 (92.026) Test: [2410/2500] Time 0.036 (0.038) Loss 1.8734 (1.0380) Acc@1 55.000 (74.019) Acc@5 80.000 (91.966) Test: [2420/2500] Time 0.036 (0.038) Loss 2.5639 (1.0400) Acc@1 35.000 (73.982) Acc@5 70.000 (91.945) Test: [2430/2500] Time 0.036 (0.038) Loss 1.1588 (1.0429) Acc@1 60.000 (73.908) Acc@5 100.000 (91.909) Test: [2440/2500] Time 0.036 (0.038) Loss 0.9905 (1.0431) Acc@1 75.000 (73.898) Acc@5 95.000 (91.930) Test: [2450/2500] Time 0.036 (0.038) Loss 0.2587 (1.0450) Acc@1 95.000 (73.837) Acc@5 95.000 (91.930) Test: [2460/2500] Time 0.036 (0.038) Loss 0.2842 (1.0445) Acc@1 90.000 (73.862) Acc@5 100.000 (91.934) Test: [2470/2500] Time 0.036 (0.038) Loss 0.3358 (1.0420) Acc@1 95.000 (73.909) Acc@5 100.000 (91.951) Test: [2480/2500] Time 0.037 (0.038) Loss 0.7659 (1.0391) Acc@1 90.000 (73.986) Acc@5 90.000 (91.973) Test: [2490/2500] Time 0.036 (0.038) Loss 0.4934 (1.0366) Acc@1 85.000 (74.045) Acc@5 100.000 (91.997)