cccjc's picture
add weights
f4fb17d
2020-09-10 12:44:43,500 Loading pretrained backbone weights from hdfs:///home/byte_arnold_hl_vc/user/chenjiacheng/data/coco/original_updown/original_updown_backbone.pth for backbone source vsepp_detector
2020-09-10 12:44:48,017 Resnet backbone now has fixed blocks 2
2020-09-10 12:44:48,223 Use adam as the optimizer, with init lr 0.0005
2020-09-10 12:44:48,223 Image encoder is data paralleled now.
2020-09-10 12:44:48,350 Load full model with backbone
2020-09-10 12:44:48,352 Loading dataset
2020-09-10 12:44:52,867 Input mode small: scaled by factor 2.0
2020-09-10 12:44:57,306 Computing results...
2020-09-10 12:45:32,345 Test: [0/196] Le 61.5342 (61.5341) Time 35.037 (0.000)
2020-09-10 12:45:35,608 Test: [10/196] Le 62.2255 (62.1526) Time 0.321 (0.000)
2020-09-10 12:45:42,461 Test: [20/196] Le 65.6731 (62.1577) Time 0.371 (0.000)
2020-09-10 12:45:48,545 Test: [30/196] Le 65.8112 (62.3998) Time 0.599 (0.000)
2020-09-10 12:45:56,936 Test: [40/196] Le 63.2921 (62.5376) Time 1.323 (0.000)
2020-09-10 12:46:03,611 Test: [50/196] Le 63.3793 (62.4683) Time 1.352 (0.000)
2020-09-10 12:46:11,003 Test: [60/196] Le 59.8511 (62.3342) Time 1.421 (0.000)
2020-09-10 12:46:17,578 Test: [70/196] Le 60.9663 (62.2750) Time 1.402 (0.000)
2020-09-10 12:46:25,291 Test: [80/196] Le 65.1833 (62.4220) Time 0.678 (0.000)
2020-09-10 12:46:31,947 Test: [90/196] Le 63.3992 (62.4493) Time 0.986 (0.000)
2020-09-10 12:46:39,390 Test: [100/196] Le 62.4223 (62.4397) Time 0.317 (0.000)
2020-09-10 12:46:45,981 Test: [110/196] Le 59.4193 (62.3932) Time 0.315 (0.000)
2020-09-10 12:46:54,739 Test: [120/196] Le 63.0641 (62.4526) Time 0.324 (0.000)
2020-09-10 12:47:00,989 Test: [130/196] Le 65.7017 (62.4476) Time 0.320 (0.000)
2020-09-10 12:47:08,321 Test: [140/196] Le 67.1813 (62.4422) Time 0.367 (0.000)
2020-09-10 12:47:15,198 Test: [150/196] Le 63.0721 (62.4588) Time 0.319 (0.000)
2020-09-10 12:47:22,794 Test: [160/196] Le 61.7248 (62.4631) Time 0.320 (0.000)
2020-09-10 12:47:29,238 Test: [170/196] Le 62.3504 (62.4660) Time 0.317 (0.000)
2020-09-10 12:47:36,362 Test: [180/196] Le 60.4507 (62.4707) Time 0.321 (0.000)
2020-09-10 12:47:42,892 Test: [190/196] Le 61.8195 (62.4209) Time 0.323 (0.000)
2020-09-10 12:47:46,943 Images: 5000, Captions: 25000
2020-09-10 12:48:20,121 calculate similarity time:
2020-09-10 12:48:20,491 Image to text: 78.8, 96.4, 98.2, 1.0, 1.7
2020-09-10 12:48:20,777 Text to image: 63.4, 91.3, 96.0, 1.0, 3.9
2020-09-10 12:48:20,777 rsum: 524.1 ar: 91.1 ari: 83.6
2020-09-10 12:48:20,855 calculate similarity time:
2020-09-10 12:48:21,213 Image to text: 79.2, 94.8, 98.3, 1.0, 1.9
2020-09-10 12:48:21,507 Text to image: 61.9, 89.8, 95.4, 1.0, 4.4
2020-09-10 12:48:21,507 rsum: 519.4 ar: 90.8 ari: 82.4
2020-09-10 12:48:21,577 calculate similarity time:
2020-09-10 12:48:21,917 Image to text: 79.4, 96.3, 99.2, 1.0, 1.6
2020-09-10 12:48:22,210 Text to image: 63.9, 90.4, 96.0, 1.0, 3.7
2020-09-10 12:48:22,210 rsum: 525.3 ar: 91.6 ari: 83.5
2020-09-10 12:48:22,304 calculate similarity time:
2020-09-10 12:48:22,646 Image to text: 75.6, 95.4, 98.0, 1.0, 2.0
2020-09-10 12:48:22,939 Text to image: 61.2, 90.1, 95.8, 1.0, 3.5
2020-09-10 12:48:22,940 rsum: 516.1 ar: 89.7 ari: 82.4
2020-09-10 12:48:23,016 calculate similarity time:
2020-09-10 12:48:23,359 Image to text: 76.9, 96.1, 98.7, 1.0, 1.8
2020-09-10 12:48:23,651 Text to image: 62.7, 91.4, 96.7, 1.0, 3.4
2020-09-10 12:48:23,651 rsum: 522.5 ar: 90.6 ari: 83.6
2020-09-10 12:48:23,651 -----------------------------------
2020-09-10 12:48:23,651 Mean metrics:
2020-09-10 12:48:23,651 rsum: 521.5
2020-09-10 12:48:23,651 Average i2t Recall: 90.8
2020-09-10 12:48:23,651 Image to text: 78.0 95.8 98.5 1.0 1.8
2020-09-10 12:48:23,651 Average t2i Recall: 83.1
2020-09-10 12:48:23,651 Text to image: 62.6 90.6 96.0 1.0 3.8
2020-09-10 12:48:27,204 Loading pretrained backbone weights from hdfs:///home/byte_arnold_hl_vc/user/chenjiacheng/data/coco/original_updown/original_updown_backbone.pth for backbone source vsepp_detector
2020-09-10 12:48:32,040 Resnet backbone now has fixed blocks 2
2020-09-10 12:48:32,213 Use adam as the optimizer, with init lr 0.0005
2020-09-10 12:48:32,213 Image encoder is data paralleled now.
2020-09-10 12:48:32,349 Load full model with backbone
2020-09-10 12:48:32,351 Loading dataset
2020-09-10 12:48:35,620 Input mode small: scaled by factor 2.0
2020-09-10 12:48:41,792 Computing results...
2020-09-10 12:48:58,912 Test: [0/196] Le 61.5342 (61.5341) Time 17.117 (0.000)
2020-09-10 12:49:05,627 Test: [10/196] Le 62.2255 (62.1526) Time 1.811 (0.000)
2020-09-10 12:49:12,470 Test: [20/196] Le 65.6731 (62.1577) Time 0.571 (0.000)
2020-09-10 12:49:18,959 Test: [30/196] Le 65.8112 (62.3998) Time 0.907 (0.000)
2020-09-10 12:49:27,170 Test: [40/196] Le 63.2921 (62.5376) Time 0.355 (0.000)
2020-09-10 12:49:33,808 Test: [50/196] Le 63.3793 (62.4683) Time 0.335 (0.000)
2020-09-10 12:49:43,444 Test: [60/196] Le 59.8511 (62.3342) Time 0.390 (0.000)
2020-09-10 12:49:48,822 Test: [70/196] Le 60.9663 (62.2750) Time 0.338 (0.000)
2020-09-10 12:49:56,979 Test: [80/196] Le 65.1833 (62.4220) Time 0.320 (0.000)
2020-09-10 12:50:03,598 Test: [90/196] Le 63.3992 (62.4493) Time 0.316 (0.000)
2020-09-10 12:50:11,072 Test: [100/196] Le 62.4223 (62.4397) Time 0.326 (0.000)
2020-09-10 12:50:17,790 Test: [110/196] Le 59.4193 (62.3932) Time 0.328 (0.000)
2020-09-10 12:50:25,012 Test: [120/196] Le 63.0641 (62.4526) Time 0.330 (0.000)
2020-09-10 12:50:31,297 Test: [130/196] Le 65.7017 (62.4476) Time 0.325 (0.000)
2020-09-10 12:50:38,239 Test: [140/196] Le 67.1813 (62.4422) Time 0.326 (0.000)
2020-09-10 12:50:45,220 Test: [150/196] Le 63.0721 (62.4588) Time 0.319 (0.000)
2020-09-10 12:50:52,620 Test: [160/196] Le 61.7248 (62.4631) Time 0.321 (0.000)
2020-09-10 12:50:59,469 Test: [170/196] Le 62.3504 (62.4660) Time 0.322 (0.000)
2020-09-10 12:51:07,511 Test: [180/196] Le 60.4507 (62.4707) Time 0.324 (0.000)
2020-09-10 12:51:14,207 Test: [190/196] Le 61.8195 (62.4209) Time 0.322 (0.000)
2020-09-10 12:51:17,881 Images: 5000, Captions: 25000
2020-09-10 12:51:45,001 calculate similarity time:
2020-09-10 12:52:03,960 rsum: 423.8
2020-09-10 12:52:03,960 Average i2t Recall: 77.0
2020-09-10 12:52:03,960 Image to text: 56.2 83.7 90.9 1.0 4.9
2020-09-10 12:52:03,960 Average t2i Recall: 64.3
2020-09-10 12:52:03,960 Text to image: 40.8 70.6 81.5 2.0 14.7