Skip to content

GitLab

Commit cea8529d authored by zhangqha@sugon.com's avatar zhangqha@sugon.com
Browse files

上传代码

parent f5e0c216
Hide whitespace changes
Inline Side-by-side
Showing with 3240 additions and 0 deletions
README.md
View file @ cea8529d
# MobileNet_v3
## 论文
Searching for MobileNetV3
https://arxiv.org/pdf/1905.02244.pdf
## 模型结构
MobileNetv3模型采用轻量级的深度可分离卷积（depthwise separable convolution）结构，以减少模型参数量和计算复杂度。
![Backbone](Backbone.png)
## 算法原理
MobileNetv3模型采用混合使用轻量级深度可分离卷积和逆残差结构（Inverted Residuals）的算法原理，以实现高效计算和良好的模型性能
![Algorithm_principle](Algorithm_principle.png)
## 环境配置
### Docker(方法一)
此处提供[光源](https://www.sourcefind.cn/#/service-list)拉取docker镜像的地址与使用步骤
```
docker pull image.sourcefind.cn:5000/dcu/admin/base/custom:tvm_fp32_dtk22.10.1_py38_centos-7.6-latest
docker run -it -v /path/your_code_data/:/path/your_code_data/ --shm-size=32G --privileged=true --device=/dev/kfd --device=/dev/dri/ --group-add video --name docker_name imageID bash
cd /root/tvm-0.11-dev0/apps/
git clone http://developer.hpccube.com/codes/modelzoo/mobilenet_v3_tvm.git
```
## 数据集
在本测试中可以使用ImageNet数据集。
下载ImageNet数据集：https://image-net.org/
下载val数据：链接：https://pan.baidu.com/s/1oXsmsYahGVG3uOZ8e535LA?pwd=c3bc 提取码：c3bc 替换ImageNet数据集中的val目录，处理后的数据结构如下：
```
data
├──imagenet
├── meta
├──val.txt
├──train.txt
...
├── train
├── val
```
## 推理
Python Deploy测试参考：
```
cd mobilenet_v3_tvm
ulimit -s unlimited
export MIOPEN_DEBUG_CONV_IMPLICIT_GEMM=0
export HIP_VISIBLE_DEVICES=2(此处修改为对应加速卡ID号)
python val_onnx.py --test-only --data-path /parastor/DL_DATA/ImageNet-pytorch --model mobilenet_v3_large --b 1 --pretrained
```
### result
![result](result.png)
## 精度
测试数据：ImageNet-pytorch，使用的加速卡:Z100L。
```
Test: [ 0/50000] eta: 21:13:14 loss: 3.7605 (3.7605) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 1.5279 data: 1.2693 max mem: 86
Test: [ 100/50000] eta: 0:16:33 loss: 0.0005 (0.5272) acc1: 100.0000 (88.1188) acc5: 100.0000 (97.0297) time: 0.0051 data: 0.0002 max mem: 86
Test: [ 200/50000] eta: 0:10:23 loss: 0.1722 (0.6195) acc1: 100.0000 (82.5871) acc5: 100.0000 (96.5174) time: 0.0050 data: 0.0002 max mem: 86
Test: [ 300/50000] eta: 0:08:11 loss: 0.0325 (0.6624) acc1: 100.0000 (82.0598) acc5: 100.0000 (96.0133) time: 0.0046 data: 0.0002 max mem: 86
Test: [ 400/50000] eta: 0:07:03 loss: 0.0283 (0.7113) acc1: 100.0000 (81.7955) acc5: 100.0000 (94.7631) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 500/50000] eta: 0:06:19 loss: 0.0000 (0.6293) acc1: 100.0000 (83.8323) acc5: 100.0000 (95.4092) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 600/50000] eta: 0:05:51 loss: 0.0002 (0.5730) acc1: 100.0000 (85.3577) acc5: 100.0000 (95.8403) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 700/50000] eta: 0:05:30 loss: 0.0023 (0.5215) acc1: 100.0000 (86.5906) acc5: 100.0000 (96.2910) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 800/50000] eta: 0:05:13 loss: 0.0003 (0.4884) acc1: 100.0000 (87.5156) acc5: 100.0000 (96.5044) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 900/50000] eta: 0:04:59 loss: 0.0004 (0.4812) acc1: 100.0000 (87.6804) acc5: 100.0000 (96.8923) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 1000/50000] eta: 0:04:50 loss: 0.0004 (0.4654) acc1: 100.0000 (88.4116) acc5: 100.0000 (97.0030) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 1100/50000] eta: 0:04:42 loss: 0.0158 (0.5022) acc1: 100.0000 (87.4659) acc5: 100.0000 (96.5486) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 1200/50000] eta: 0:04:35 loss: 0.0051 (0.4699) acc1: 100.0000 (88.1765) acc5: 100.0000 (96.8360) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 1300/50000] eta: 0:04:30 loss: 0.0049 (0.4499) acc1: 100.0000 (88.8547) acc5: 100.0000 (96.9254) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 1400/50000] eta: 0:04:24 loss: 0.0190 (0.5285) acc1: 100.0000 (87.2234) acc5: 100.0000 (96.4311) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 1500/50000] eta: 0:04:19 loss: 0.0029 (0.5403) acc1: 100.0000 (87.1419) acc5: 100.0000 (96.1359) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 1600/50000] eta: 0:04:16 loss: 0.0291 (0.5636) acc1: 100.0000 (86.8832) acc5: 100.0000 (96.0025) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 1700/50000] eta: 0:04:13 loss: 0.0790 (0.6317) acc1: 100.0000 (85.3028) acc5: 100.0000 (95.7084) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 1800/50000] eta: 0:04:10 loss: 0.8301 (0.6861) acc1: 0.0000 (83.8978) acc5: 100.0000 (95.2804) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 1900/50000] eta: 0:04:07 loss: 0.1593 (0.6977) acc1: 100.0000 (83.2194) acc5: 100.0000 (95.4235) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 2000/50000] eta: 0:04:05 loss: 0.0165 (0.7040) acc1: 100.0000 (83.2084) acc5: 100.0000 (95.4023) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 2100/50000] eta: 0:04:03 loss: 0.0156 (0.7113) acc1: 100.0000 (82.8653) acc5: 100.0000 (95.4307) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 2200/50000] eta: 0:04:01 loss: 0.0035 (0.7085) acc1: 100.0000 (82.9623) acc5: 100.0000 (95.4112) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 2300/50000] eta: 0:03:59 loss: 0.0000 (0.7477) acc1: 100.0000 (82.3990) acc5: 100.0000 (95.0456) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 2400/50000] eta: 0:03:57 loss: 0.0144 (0.7674) acc1: 100.0000 (81.7576) acc5: 100.0000 (94.9188) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 2500/50000] eta: 0:03:55 loss: 0.1349 (0.7727) acc1: 100.0000 (81.7673) acc5: 100.0000 (94.7621) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 2600/50000] eta: 0:03:53 loss: 0.0001 (0.7679) acc1: 100.0000 (81.8916) acc5: 100.0000 (94.6943) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 2700/50000] eta: 0:03:50 loss: 0.2303 (0.7688) acc1: 100.0000 (81.7845) acc5: 100.0000 (94.6686) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 2800/50000] eta: 0:03:48 loss: 0.1836 (0.7950) acc1: 100.0000 (81.1853) acc5: 100.0000 (94.5734) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 2900/50000] eta: 0:03:47 loss: 0.0785 (0.7940) acc1: 100.0000 (81.0410) acc5: 100.0000 (94.5881) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 3000/50000] eta: 0:03:45 loss: 0.8794 (0.8190) acc1: 100.0000 (80.5398) acc5: 100.0000 (94.3019) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 3100/50000] eta: 0:03:43 loss: 0.1602 (0.8417) acc1: 100.0000 (79.8130) acc5: 100.0000 (94.3567) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 3200/50000] eta: 0:03:41 loss: 0.0120 (0.8620) acc1: 100.0000 (79.2565) acc5: 100.0000 (94.0331) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 3300/50000] eta: 0:03:40 loss: 0.1979 (0.8830) acc1: 100.0000 (78.8549) acc5: 100.0000 (93.9109) time: 0.0041 data: 0.0003 max mem: 86
Test: [ 3400/50000] eta: 0:03:39 loss: 0.8700 (0.8987) acc1: 0.0000 (78.3593) acc5: 100.0000 (93.7959) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 3500/50000] eta: 0:03:38 loss: 0.0001 (0.9023) acc1: 100.0000 (77.8920) acc5: 100.0000 (93.7446) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 3600/50000] eta: 0:03:37 loss: 0.0002 (0.8921) acc1: 100.0000 (78.0894) acc5: 100.0000 (93.7795) time: 0.0048 data: 0.0002 max mem: 86
Test: [ 3700/50000] eta: 0:03:36 loss: 1.0290 (0.8984) acc1: 0.0000 (77.5736) acc5: 100.0000 (93.8665) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 3800/50000] eta: 0:03:35 loss: 0.0186 (0.9037) acc1: 100.0000 (77.2691) acc5: 100.0000 (93.9227) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 3900/50000] eta: 0:03:35 loss: 0.0721 (0.9016) acc1: 100.0000 (77.3135) acc5: 100.0000 (93.9503) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 4000/50000] eta: 0:03:34 loss: 0.0351 (0.9084) acc1: 100.0000 (77.3057) acc5: 100.0000 (93.8765) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 4100/50000] eta: 0:03:33 loss: 0.0029 (0.8997) acc1: 100.0000 (77.4689) acc5: 100.0000 (94.0015) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 4200/50000] eta: 0:03:32 loss: 0.0013 (0.8858) acc1: 100.0000 (77.7910) acc5: 100.0000 (94.1204) time: 0.0048 data: 0.0002 max mem: 86
Test: [ 4300/50000] eta: 0:03:32 loss: 0.0011 (0.8741) acc1: 100.0000 (78.1214) acc5: 100.0000 (94.1641) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 4400/50000] eta: 0:03:31 loss: 0.0000 (0.8666) acc1: 100.0000 (78.1641) acc5: 100.0000 (94.2513) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 4500/50000] eta: 0:03:30 loss: 0.0001 (0.8528) acc1: 100.0000 (78.5603) acc5: 100.0000 (94.3568) time: 0.0046 data: 0.0002 max mem: 86
Test: [ 4600/50000] eta: 0:03:30 loss: 0.0002 (0.8414) acc1: 100.0000 (78.8524) acc5: 100.0000 (94.3708) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 4700/50000] eta: 0:03:29 loss: 0.0031 (0.8295) acc1: 100.0000 (79.1534) acc5: 100.0000 (94.4905) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 4800/50000] eta: 0:03:29 loss: 0.0007 (0.8202) acc1: 100.0000 (79.4210) acc5: 100.0000 (94.5220) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 4900/50000] eta: 0:03:28 loss: 0.0011 (0.8136) acc1: 100.0000 (79.6164) acc5: 100.0000 (94.5725) time: 0.0047 data: 0.0002 max mem: 86
Test: [ 5000/50000] eta: 0:03:27 loss: 0.0025 (0.8080) acc1: 100.0000 (79.7041) acc5: 100.0000 (94.6411) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 5100/50000] eta: 0:03:26 loss: 0.3220 (0.8063) acc1: 100.0000 (79.7099) acc5: 100.0000 (94.6481) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 5200/50000] eta: 0:03:25 loss: 0.0062 (0.8012) acc1: 100.0000 (79.8500) acc5: 100.0000 (94.6164) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 5300/50000] eta: 0:03:25 loss: 0.0002 (0.7945) acc1: 100.0000 (80.0415) acc5: 100.0000 (94.6237) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 5400/50000] eta: 0:03:24 loss: 0.0015 (0.7994) acc1: 100.0000 (80.1889) acc5: 100.0000 (94.5751) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 5500/50000] eta: 0:03:24 loss: 0.0039 (0.8008) acc1: 100.0000 (80.2218) acc5: 100.0000 (94.5646) time: 0.0046 data: 0.0002 max mem: 86
Test: [ 5600/50000] eta: 0:03:23 loss: 0.0009 (0.7973) acc1: 100.0000 (80.3428) acc5: 100.0000 (94.5903) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 5700/50000] eta: 0:03:22 loss: 0.0045 (0.8016) acc1: 100.0000 (80.3368) acc5: 100.0000 (94.4922) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 5800/50000] eta: 0:03:21 loss: 0.0441 (0.8010) acc1: 100.0000 (80.4172) acc5: 100.0000 (94.4665) time: 0.0043 data: 0.0003 max mem: 86
Test: [ 5900/50000] eta: 0:03:20 loss: 0.0016 (0.8022) acc1: 100.0000 (80.5287) acc5: 100.0000 (94.4416) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 6000/50000] eta: 0:03:20 loss: 0.0303 (0.8104) acc1: 100.0000 (80.4199) acc5: 100.0000 (94.4176) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 6100/50000] eta: 0:03:19 loss: 0.0044 (0.8116) acc1: 100.0000 (80.3475) acc5: 100.0000 (94.4271) time: 0.0038 data: 0.0001 max mem: 86
Test: [ 6200/50000] eta: 0:03:18 loss: 0.0818 (0.8107) acc1: 100.0000 (80.3580) acc5: 100.0000 (94.3719) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 6300/50000] eta: 0:03:17 loss: 0.0063 (0.8139) acc1: 100.0000 (80.2888) acc5: 100.0000 (94.3977) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 6400/50000] eta: 0:03:17 loss: 0.0055 (0.8195) acc1: 100.0000 (80.3000) acc5: 100.0000 (94.2978) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 6500/50000] eta: 0:03:16 loss: 0.0001 (0.8117) acc1: 100.0000 (80.4645) acc5: 100.0000 (94.3393) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 6600/50000] eta: 0:03:15 loss: 0.0092 (0.8027) acc1: 100.0000 (80.6696) acc5: 100.0000 (94.4251) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 6700/50000] eta: 0:03:15 loss: 0.0012 (0.7978) acc1: 100.0000 (80.7790) acc5: 100.0000 (94.4337) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 6800/50000] eta: 0:03:14 loss: 0.0000 (0.7927) acc1: 100.0000 (80.8558) acc5: 100.0000 (94.4861) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 6900/50000] eta: 0:03:13 loss: 0.0122 (0.7864) acc1: 100.0000 (81.0172) acc5: 100.0000 (94.5225) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7000/50000] eta: 0:03:12 loss: 0.0003 (0.7786) acc1: 100.0000 (81.2170) acc5: 100.0000 (94.5579) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 7100/50000] eta: 0:03:11 loss: 0.0009 (0.7743) acc1: 100.0000 (81.3407) acc5: 100.0000 (94.5782) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7200/50000] eta: 0:03:11 loss: 0.0001 (0.7686) acc1: 100.0000 (81.4331) acc5: 100.0000 (94.6257) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 7300/50000] eta: 0:03:10 loss: 0.0002 (0.7626) acc1: 100.0000 (81.6053) acc5: 100.0000 (94.6583) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 7400/50000] eta: 0:03:09 loss: 0.0007 (0.7567) acc1: 100.0000 (81.7457) acc5: 100.0000 (94.7034) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7500/50000] eta: 0:03:10 loss: 0.0025 (0.7521) acc1: 100.0000 (81.8558) acc5: 100.0000 (94.7074) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7600/50000] eta: 0:03:09 loss: 0.0618 (0.7581) acc1: 100.0000 (81.7392) acc5: 100.0000 (94.6718) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7700/50000] eta: 0:03:08 loss: 0.2603 (0.7584) acc1: 100.0000 (81.7167) acc5: 100.0000 (94.6760) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7800/50000] eta: 0:03:08 loss: 0.3902 (0.7611) acc1: 100.0000 (81.6177) acc5: 100.0000 (94.6545) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 7900/50000] eta: 0:03:08 loss: 0.0121 (0.7564) acc1: 100.0000 (81.7745) acc5: 100.0000 (94.6969) time: 0.0047 data: 0.0002 max mem: 86
Test: [ 8000/50000] eta: 0:03:07 loss: 0.6661 (0.7616) acc1: 0.0000 (81.5523) acc5: 100.0000 (94.6882) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8100/50000] eta: 0:03:07 loss: 0.0326 (0.7601) acc1: 100.0000 (81.5578) acc5: 100.0000 (94.6797) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 8200/50000] eta: 0:03:07 loss: 0.6533 (0.7703) acc1: 100.0000 (81.4413) acc5: 100.0000 (94.5982) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8300/50000] eta: 0:03:06 loss: 0.5622 (0.7768) acc1: 100.0000 (81.3637) acc5: 100.0000 (94.5428) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8400/50000] eta: 0:03:06 loss: 0.9667 (0.7885) acc1: 0.0000 (80.9666) acc5: 100.0000 (94.4768) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 8500/50000] eta: 0:03:06 loss: 0.0256 (0.7869) acc1: 100.0000 (81.0022) acc5: 100.0000 (94.5065) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8600/50000] eta: 0:03:05 loss: 0.0389 (0.7950) acc1: 100.0000 (80.8394) acc5: 100.0000 (94.4774) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 8700/50000] eta: 0:03:05 loss: 0.2748 (0.7973) acc1: 100.0000 (80.7378) acc5: 100.0000 (94.4604) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8800/50000] eta: 0:03:04 loss: 0.1653 (0.8007) acc1: 100.0000 (80.7181) acc5: 100.0000 (94.4438) time: 0.0046 data: 0.0003 max mem: 86
Test: [ 8900/50000] eta: 0:03:04 loss: 0.1470 (0.7986) acc1: 100.0000 (80.7550) acc5: 100.0000 (94.4388) time: 0.0046 data: 0.0002 max mem: 86
Test: [ 9000/50000] eta: 0:03:03 loss: 0.2561 (0.7993) acc1: 100.0000 (80.7133) acc5: 100.0000 (94.4340) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 9100/50000] eta: 0:03:03 loss: 0.0316 (0.7989) acc1: 100.0000 (80.6725) acc5: 100.0000 (94.4621) time: 0.0036 data: 0.0002 max mem: 86
Test: [ 9200/50000] eta: 0:03:02 loss: 0.3239 (0.7978) acc1: 100.0000 (80.6543) acc5: 100.0000 (94.4680) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 9300/50000] eta: 0:03:01 loss: 0.1094 (0.8015) acc1: 100.0000 (80.5075) acc5: 100.0000 (94.4737) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 9400/50000] eta: 0:03:00 loss: 0.2506 (0.8040) acc1: 100.0000 (80.4489) acc5: 100.0000 (94.4580) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 9500/50000] eta: 0:03:00 loss: 0.1578 (0.8035) acc1: 100.0000 (80.4021) acc5: 100.0000 (94.4743) time: 0.0047 data: 0.0002 max mem: 86
Test: [ 9600/50000] eta: 0:03:00 loss: 0.0188 (0.8056) acc1: 100.0000 (80.4187) acc5: 100.0000 (94.4381) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 9700/50000] eta: 0:02:59 loss: 0.4514 (0.8102) acc1: 100.0000 (80.2495) acc5: 100.0000 (94.4233) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 9800/50000] eta: 0:02:58 loss: 0.0612 (0.8095) acc1: 100.0000 (80.2775) acc5: 100.0000 (94.4189) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 9900/50000] eta: 0:02:58 loss: 0.3089 (0.8120) acc1: 100.0000 (80.2242) acc5: 100.0000 (94.4248) time: 0.0043 data: 0.0002 max mem: 86
Test: [10000/50000] eta: 0:02:57 loss: 0.0415 (0.8116) acc1: 100.0000 (80.2320) acc5: 100.0000 (94.4606) time: 0.0041 data: 0.0002 max mem: 86
Test: [10100/50000] eta: 0:02:57 loss: 0.5171 (0.8145) acc1: 100.0000 (80.0614) acc5: 100.0000 (94.4758) time: 0.0040 data: 0.0002 max mem: 86
Test: [10200/50000] eta: 0:02:56 loss: 0.0181 (0.8114) acc1: 100.0000 (80.0902) acc5: 100.0000 (94.5103) time: 0.0041 data: 0.0002 max mem: 86
Test: [10300/50000] eta: 0:02:56 loss: 0.0997 (0.8119) acc1: 100.0000 (80.0408) acc5: 100.0000 (94.5345) time: 0.0041 data: 0.0002 max mem: 86
Test: [10400/50000] eta: 0:02:55 loss: 0.0290 (0.8098) acc1: 100.0000 (80.0788) acc5: 100.0000 (94.5582) time: 0.0041 data: 0.0002 max mem: 86
Test: [10500/50000] eta: 0:02:55 loss: 0.0330 (0.8097) acc1: 100.0000 (80.1257) acc5: 100.0000 (94.5624) time: 0.0041 data: 0.0002 max mem: 86
Test: [10600/50000] eta: 0:02:54 loss: 0.1191 (0.8095) acc1: 100.0000 (80.1057) acc5: 100.0000 (94.5477) time: 0.0042 data: 0.0002 max mem: 86
Test: [10700/50000] eta: 0:02:54 loss: 0.0099 (0.8100) acc1: 100.0000 (80.0953) acc5: 100.0000 (94.5426) time: 0.0042 data: 0.0002 max mem: 86
Test: [10800/50000] eta: 0:02:53 loss: 0.2072 (0.8087) acc1: 100.0000 (80.1315) acc5: 100.0000 (94.5468) time: 0.0043 data: 0.0002 max mem: 86
Test: [10900/50000] eta: 0:02:53 loss: 0.0128 (0.8042) acc1: 100.0000 (80.2312) acc5: 100.0000 (94.5785) time: 0.0047 data: 0.0002 max mem: 86
Test: [11000/50000] eta: 0:02:52 loss: 0.0486 (0.8027) acc1: 100.0000 (80.2382) acc5: 100.0000 (94.6187) time: 0.0042 data: 0.0002 max mem: 86
Test: [11100/50000] eta: 0:02:52 loss: 0.0621 (0.8054) acc1: 100.0000 (80.1910) acc5: 100.0000 (94.6041) time: 0.0042 data: 0.0002 max mem: 86
Test: [11200/50000] eta: 0:02:51 loss: 0.0130 (0.8056) acc1: 100.0000 (80.1893) acc5: 100.0000 (94.5719) time: 0.0042 data: 0.0002 max mem: 86
Test: [11300/50000] eta: 0:02:51 loss: 0.0220 (0.8028) acc1: 100.0000 (80.2407) acc5: 100.0000 (94.5934) time: 0.0042 data: 0.0002 max mem: 86
Test: [11400/50000] eta: 0:02:50 loss: 0.0704 (0.8124) acc1: 100.0000 (80.1333) acc5: 100.0000 (94.5180) time: 0.0042 data: 0.0002 max mem: 86
Test: [11500/50000] eta: 0:02:50 loss: 0.0229 (0.8090) acc1: 100.0000 (80.2278) acc5: 100.0000 (94.5135) time: 0.0043 data: 0.0002 max mem: 86
Test: [11600/50000] eta: 0:02:49 loss: 0.3336 (0.8100) acc1: 100.0000 (80.1396) acc5: 100.0000 (94.5005) time: 0.0044 data: 0.0002 max mem: 86
Test: [11700/50000] eta: 0:02:49 loss: 0.0793 (0.8110) acc1: 100.0000 (80.1214) acc5: 100.0000 (94.4877) time: 0.0042 data: 0.0002 max mem: 86
Test: [11800/50000] eta: 0:02:48 loss: 0.0249 (0.8095) acc1: 100.0000 (80.1542) acc5: 100.0000 (94.5174) time: 0.0040 data: 0.0002 max mem: 86
Test: [11900/50000] eta: 0:02:47 loss: 0.0822 (0.8103) acc1: 100.0000 (80.1361) acc5: 100.0000 (94.4711) time: 0.0042 data: 0.0002 max mem: 86
Test: [12000/50000] eta: 0:02:47 loss: 0.0385 (0.8100) acc1: 100.0000 (80.1350) acc5: 100.0000 (94.5088) time: 0.0041 data: 0.0002 max mem: 86
Test: [12100/50000] eta: 0:02:46 loss: 0.2712 (0.8136) acc1: 100.0000 (79.9273) acc5: 100.0000 (94.5376) time: 0.0041 data: 0.0002 max mem: 86
Test: [12200/50000] eta: 0:02:46 loss: 0.0601 (0.8135) acc1: 100.0000 (79.8951) acc5: 100.0000 (94.5496) time: 0.0040 data: 0.0002 max mem: 86
Test: [12300/50000] eta: 0:02:45 loss: 0.0080 (0.8108) acc1: 100.0000 (79.9610) acc5: 100.0000 (94.5695) time: 0.0043 data: 0.0002 max mem: 86
Test: [12400/50000] eta: 0:02:45 loss: 0.0194 (0.8125) acc1: 100.0000 (79.9129) acc5: 100.0000 (94.5488) time: 0.0044 data: 0.0002 max mem: 86
Test: [12500/50000] eta: 0:02:44 loss: 0.2914 (0.8163) acc1: 100.0000 (79.7696) acc5: 100.0000 (94.5444) time: 0.0042 data: 0.0002 max mem: 86
Test: [12600/50000] eta: 0:02:44 loss: 0.0031 (0.8145) acc1: 100.0000 (79.7794) acc5: 100.0000 (94.5877) time: 0.0040 data: 0.0002 max mem: 86
Test: [12700/50000] eta: 0:02:43 loss: 0.0760 (0.8128) acc1: 100.0000 (79.7969) acc5: 100.0000 (94.6146) time: 0.0039 data: 0.0002 max mem: 86
Test: [12800/50000] eta: 0:02:43 loss: 0.0001 (0.8072) acc1: 100.0000 (79.9234) acc5: 100.0000 (94.6489) time: 0.0038 data: 0.0002 max mem: 86
Test: [12900/50000] eta: 0:02:42 loss: 0.0351 (0.8071) acc1: 100.0000 (79.9008) acc5: 100.0000 (94.6516) time: 0.0040 data: 0.0002 max mem: 86
Test: [13000/50000] eta: 0:02:42 loss: 0.0030 (0.8019) acc1: 100.0000 (80.0323) acc5: 100.0000 (94.6927) time: 0.0040 data: 0.0002 max mem: 86
Test: [13100/50000] eta: 0:02:41 loss: 0.0005 (0.7990) acc1: 100.0000 (80.1084) acc5: 100.0000 (94.7180) time: 0.0039 data: 0.0002 max mem: 86
Test: [13200/50000] eta: 0:02:40 loss: 0.0728 (0.7972) acc1: 100.0000 (80.1454) acc5: 100.0000 (94.7352) time: 0.0037 data: 0.0002 max mem: 86
Test: [13300/50000] eta: 0:02:40 loss: 0.6036 (0.7994) acc1: 100.0000 (80.0241) acc5: 100.0000 (94.7372) time: 0.0037 data: 0.0002 max mem: 86
Test: [13400/50000] eta: 0:02:39 loss: 0.0842 (0.8006) acc1: 100.0000 (79.9269) acc5: 100.0000 (94.7392) time: 0.0050 data: 0.0002 max mem: 86
Test: [13500/50000] eta: 0:02:39 loss: 0.3301 (0.7995) acc1: 100.0000 (79.9348) acc5: 100.0000 (94.7634) time: 0.0039 data: 0.0002 max mem: 86
Test: [13600/50000] eta: 0:02:38 loss: 0.4816 (0.8014) acc1: 100.0000 (79.8765) acc5: 100.0000 (94.7651) time: 0.0039 data: 0.0001 max mem: 86
Test: [13700/50000] eta: 0:02:38 loss: 0.0488 (0.8076) acc1: 100.0000 (79.8044) acc5: 100.0000 (94.7376) time: 0.0042 data: 0.0002 max mem: 86
Test: [13800/50000] eta: 0:02:37 loss: 0.0009 (0.8044) acc1: 100.0000 (79.9145) acc5: 100.0000 (94.7685) time: 0.0037 data: 0.0002 max mem: 86
Test: [13900/50000] eta: 0:02:37 loss: 0.2700 (0.8054) acc1: 100.0000 (79.8576) acc5: 100.0000 (94.7917) time: 0.0040 data: 0.0002 max mem: 86
Test: [14000/50000] eta: 0:02:36 loss: 0.0046 (0.8038) acc1: 100.0000 (79.8657) acc5: 100.0000 (94.8218) time: 0.0038 data: 0.0002 max mem: 86
Test: [14100/50000] eta: 0:02:36 loss: 0.6475 (0.8075) acc1: 100.0000 (79.7603) acc5: 100.0000 (94.8089) time: 0.0049 data: 0.0002 max mem: 86
Test: [14200/50000] eta: 0:02:35 loss: 0.0036 (0.8107) acc1: 100.0000 (79.5930) acc5: 100.0000 (94.8032) time: 0.0036 data: 0.0002 max mem: 86
Test: [14300/50000] eta: 0:02:34 loss: 0.8769 (0.8150) acc1: 0.0000 (79.5189) acc5: 100.0000 (94.7416) time: 0.0038 data: 0.0003 max mem: 86
Test: [14400/50000] eta: 0:02:34 loss: 0.0047 (0.8161) acc1: 100.0000 (79.5500) acc5: 100.0000 (94.7365) time: 0.0037 data: 0.0002 max mem: 86
Test: [14500/50000] eta: 0:02:33 loss: 0.0571 (0.8135) acc1: 100.0000 (79.5945) acc5: 100.0000 (94.7659) time: 0.0038 data: 0.0002 max mem: 86
Test: [14600/50000] eta: 0:02:33 loss: 0.0004 (0.8125) acc1: 100.0000 (79.6384) acc5: 100.0000 (94.7743) time: 0.0035 data: 0.0002 max mem: 86
Test: [14700/50000] eta: 0:02:32 loss: 0.0237 (0.8089) acc1: 100.0000 (79.7293) acc5: 100.0000 (94.8031) time: 0.0099 data: 0.0060 max mem: 86
Test: [14800/50000] eta: 0:02:32 loss: 0.0363 (0.8052) acc1: 100.0000 (79.7784) acc5: 100.0000 (94.8382) time: 0.0038 data: 0.0002 max mem: 86
Test: [14900/50000] eta: 0:02:31 loss: 0.0038 (0.8080) acc1: 100.0000 (79.7933) acc5: 100.0000 (94.8259) time: 0.0041 data: 0.0002 max mem: 86
Test: [15000/50000] eta: 0:02:31 loss: 0.1103 (0.8090) acc1: 100.0000 (79.7680) acc5: 100.0000 (94.8203) time: 0.0050 data: 0.0002 max mem: 86
Test: [15100/50000] eta: 0:02:31 loss: 0.0379 (0.8061) acc1: 100.0000 (79.8027) acc5: 100.0000 (94.8414) time: 0.0043 data: 0.0002 max mem: 86
Test: [15200/50000] eta: 0:02:30 loss: 0.2943 (0.8078) acc1: 100.0000 (79.7448) acc5: 100.0000 (94.8622) time: 0.0042 data: 0.0002 max mem: 86
Test: [15300/50000] eta: 0:02:30 loss: 0.0289 (0.8078) acc1: 100.0000 (79.7137) acc5: 100.0000 (94.8827) time: 0.0045 data: 0.0002 max mem: 86
Test: [15400/50000] eta: 0:02:29 loss: 0.0153 (0.8058) acc1: 100.0000 (79.7546) acc5: 100.0000 (94.8964) time: 0.0042 data: 0.0002 max mem: 86
Test: [15500/50000] eta: 0:02:29 loss: 0.0358 (0.8057) acc1: 100.0000 (79.7239) acc5: 100.0000 (94.8907) time: 0.0038 data: 0.0002 max mem: 86
Test: [15600/50000] eta: 0:02:28 loss: 0.1787 (0.8104) acc1: 100.0000 (79.6103) acc5: 100.0000 (94.8529) time: 0.0039 data: 0.0002 max mem: 86
Test: [15700/50000] eta: 0:02:28 loss: 0.2585 (0.8120) acc1: 100.0000 (79.5172) acc5: 100.0000 (94.8602) time: 0.0040 data: 0.0002 max mem: 86
Test: [15800/50000] eta: 0:02:27 loss: 0.0670 (0.8148) acc1: 100.0000 (79.5140) acc5: 100.0000 (94.8168) time: 0.0038 data: 0.0002 max mem: 86
Test: [15900/50000] eta: 0:02:27 loss: 0.0272 (0.8127) acc1: 100.0000 (79.5736) acc5: 100.0000 (94.8368) time: 0.0039 data: 0.0002 max mem: 86
Test: [16000/50000] eta: 0:02:26 loss: 0.0547 (0.8122) acc1: 100.0000 (79.6013) acc5: 100.0000 (94.8503) time: 0.0041 data: 0.0002 max mem: 86
Test: [16100/50000] eta: 0:02:26 loss: 0.0000 (0.8089) acc1: 100.0000 (79.6907) acc5: 100.0000 (94.8699) time: 0.0041 data: 0.0002 max mem: 86
Test: [16200/50000] eta: 0:02:25 loss: 0.0006 (0.8054) acc1: 100.0000 (79.7852) acc5: 100.0000 (94.8954) time: 0.0040 data: 0.0002 max mem: 86
Test: [16300/50000] eta: 0:02:25 loss: 0.0003 (0.8019) acc1: 100.0000 (79.8724) acc5: 100.0000 (94.9267) time: 0.0039 data: 0.0002 max mem: 86
Test: [16400/50000] eta: 0:02:24 loss: 0.0062 (0.8004) acc1: 100.0000 (79.9281) acc5: 100.0000 (94.9332) time: 0.0041 data: 0.0002 max mem: 86
Test: [16500/50000] eta: 0:02:24 loss: 0.0020 (0.8020) acc1: 100.0000 (79.9224) acc5: 100.0000 (94.8973) time: 0.0039 data: 0.0002 max mem: 86
Test: [16600/50000] eta: 0:02:24 loss: 0.0321 (0.8014) acc1: 100.0000 (79.9108) acc5: 100.0000 (94.9160) time: 0.0043 data: 0.0002 max mem: 86
Test: [16700/50000] eta: 0:02:23 loss: 0.0001 (0.7972) acc1: 100.0000 (80.0132) acc5: 100.0000 (94.9464) time: 0.0043 data: 0.0002 max mem: 86
Test: [16800/50000] eta: 0:02:23 loss: 0.0031 (0.7957) acc1: 100.0000 (80.0488) acc5: 100.0000 (94.9646) time: 0.0044 data: 0.0002 max mem: 86
Test: [16900/50000] eta: 0:02:22 loss: 0.0138 (0.7970) acc1: 100.0000 (80.0781) acc5: 100.0000 (94.9707) time: 0.0043 data: 0.0002 max mem: 86
Test: [17000/50000] eta: 0:02:22 loss: 0.0021 (0.7951) acc1: 100.0000 (80.1306) acc5: 100.0000 (94.9768) time: 0.0040 data: 0.0002 max mem: 86
Test: [17100/50000] eta: 0:02:21 loss: 1.9055 (0.7958) acc1: 0.0000 (80.1181) acc5: 100.0000 (94.9594) time: 0.0038 data: 0.0002 max mem: 86
Test: [17200/50000] eta: 0:02:21 loss: 0.0138 (0.7940) acc1: 100.0000 (80.1523) acc5: 100.0000 (94.9770) time: 0.0041 data: 0.0002 max mem: 86
Test: [17300/50000] eta: 0:02:20 loss: 0.4505 (0.7939) acc1: 100.0000 (80.1225) acc5: 100.0000 (94.9829) time: 0.0044 data: 0.0002 max mem: 86
Test: [17400/50000] eta: 0:02:20 loss: 0.0002 (0.7931) acc1: 100.0000 (80.1448) acc5: 100.0000 (94.9601) time: 0.0038 data: 0.0002 max mem: 86
Test: [17500/50000] eta: 0:02:19 loss: 0.3345 (0.7926) acc1: 100.0000 (80.0640) acc5: 100.0000 (94.9774) time: 0.0039 data: 0.0002 max mem: 86
Test: [17600/50000] eta: 0:02:19 loss: 0.0260 (0.7892) acc1: 100.0000 (80.1432) acc5: 100.0000 (95.0003) time: 0.0039 data: 0.0003 max mem: 86
Test: [17700/50000] eta: 0:02:18 loss: 0.0452 (0.7887) acc1: 100.0000 (80.1198) acc5: 100.0000 (95.0172) time: 0.0038 data: 0.0002 max mem: 86
Test: [17800/50000] eta: 0:02:18 loss: 0.0026 (0.7865) acc1: 100.0000 (80.1865) acc5: 100.0000 (95.0171) time: 0.0037 data: 0.0002 max mem: 86
Test: [17900/50000] eta: 0:02:17 loss: 0.0605 (0.7897) acc1: 100.0000 (80.1017) acc5: 100.0000 (95.0226) time: 0.0037 data: 0.0002 max mem: 86
Test: [18000/50000] eta: 0:02:17 loss: 0.1980 (0.7939) acc1: 100.0000 (79.9956) acc5: 100.0000 (95.0114) time: 0.0038 data: 0.0002 max mem: 86
Test: [18100/50000] eta: 0:02:17 loss: 0.0005 (0.7933) acc1: 100.0000 (79.9845) acc5: 100.0000 (95.0113) time: 0.0039 data: 0.0002 max mem: 86
Test: [18200/50000] eta: 0:02:16 loss: 0.0000 (0.7923) acc1: 100.0000 (80.0231) acc5: 100.0000 (95.0113) time: 0.0042 data: 0.0002 max mem: 86
Test: [18300/50000] eta: 0:02:16 loss: 0.0105 (0.7899) acc1: 100.0000 (80.1049) acc5: 100.0000 (95.0221) time: 0.0041 data: 0.0002 max mem: 86
Test: [18400/50000] eta: 0:02:15 loss: 0.0868 (0.7898) acc1: 100.0000 (80.1261) acc5: 100.0000 (95.0166) time: 0.0037 data: 0.0002 max mem: 86
Test: [18500/50000] eta: 0:02:15 loss: 0.0747 (0.7911) acc1: 100.0000 (80.1146) acc5: 100.0000 (94.9949) time: 0.0037 data: 0.0002 max mem: 86
Test: [18600/50000] eta: 0:02:14 loss: 0.0352 (0.7937) acc1: 100.0000 (80.1086) acc5: 100.0000 (94.9734) time: 0.0037 data: 0.0002 max mem: 86
Test: [18700/50000] eta: 0:02:14 loss: 0.0732 (0.7940) acc1: 100.0000 (80.1294) acc5: 100.0000 (94.9575) time: 0.0039 data: 0.0002 max mem: 86
Test: [18800/50000] eta: 0:02:13 loss: 0.0474 (0.7955) acc1: 100.0000 (80.1074) acc5: 100.0000 (94.9418) time: 0.0037 data: 0.0002 max mem: 86
Test: [18900/50000] eta: 0:02:13 loss: 0.1086 (0.7941) acc1: 100.0000 (80.1386) acc5: 100.0000 (94.9474) time: 0.0038 data: 0.0002 max mem: 86
Test: [19000/50000] eta: 0:02:12 loss: 0.0733 (0.7935) acc1: 100.0000 (80.1484) acc5: 100.0000 (94.9529) time: 0.0039 data: 0.0002 max mem: 86
Test: [19100/50000] eta: 0:02:12 loss: 1.5252 (0.7983) acc1: 0.0000 (80.0063) acc5: 100.0000 (94.9322) time: 0.0037 data: 0.0002 max mem: 86
Test: [19200/50000] eta: 0:02:11 loss: 0.0863 (0.8004) acc1: 100.0000 (79.9385) acc5: 100.0000 (94.9273) time: 0.0041 data: 0.0002 max mem: 86
Test: [19300/50000] eta: 0:02:11 loss: 0.7411 (0.7994) acc1: 100.0000 (79.9285) acc5: 100.0000 (94.9277) time: 0.0042 data: 0.0002 max mem: 86
Test: [19400/50000] eta: 0:02:10 loss: 0.0001 (0.8005) acc1: 100.0000 (79.9083) acc5: 100.0000 (94.9281) time: 0.0042 data: 0.0002 max mem: 86
Test: [19500/50000] eta: 0:02:10 loss: 0.0033 (0.8001) acc1: 100.0000 (79.9190) acc5: 100.0000 (94.9182) time: 0.0037 data: 0.0002 max mem: 86
Test: [19600/50000] eta: 0:02:09 loss: 0.0073 (0.8024) acc1: 100.0000 (79.8786) acc5: 100.0000 (94.9033) time: 0.0038 data: 0.0002 max mem: 86
Test: [19700/50000] eta: 0:02:09 loss: 0.0179 (0.8002) acc1: 100.0000 (79.9452) acc5: 100.0000 (94.9140) time: 0.0038 data: 0.0002 max mem: 86
Test: [19800/50000] eta: 0:02:08 loss: 0.1315 (0.8002) acc1: 100.0000 (79.9303) acc5: 100.0000 (94.9043) time: 0.0038 data: 0.0002 max mem: 86
Test: [19900/50000] eta: 0:02:08 loss: 0.0081 (0.7988) acc1: 100.0000 (79.9910) acc5: 100.0000 (94.9148) time: 0.0037 data: 0.0002 max mem: 86
Test: [20000/50000] eta: 0:02:07 loss: 0.1578 (0.7995) acc1: 100.0000 (79.9760) acc5: 100.0000 (94.9053) time: 0.0036 data: 0.0002 max mem: 86
Test: [20100/50000] eta: 0:02:07 loss: 0.1097 (0.8019) acc1: 100.0000 (79.9015) acc5: 100.0000 (94.8858) time: 0.0040 data: 0.0002 max mem: 86
Test: [20200/50000] eta: 0:02:06 loss: 0.0078 (0.8048) acc1: 100.0000 (79.8673) acc5: 100.0000 (94.8567) time: 0.0039 data: 0.0002 max mem: 86
Test: [20300/50000] eta: 0:02:06 loss: 0.0131 (0.8054) acc1: 100.0000 (79.8778) acc5: 100.0000 (94.8278) time: 0.0039 data: 0.0002 max mem: 86
Test: [20400/50000] eta: 0:02:05 loss: 0.0012 (0.8049) acc1: 100.0000 (79.8882) acc5: 100.0000 (94.8434) time: 0.0036 data: 0.0002 max mem: 86
Test: [20500/50000] eta: 0:02:05 loss: 0.3294 (0.8079) acc1: 100.0000 (79.8303) acc5: 100.0000 (94.8490) time: 0.0038 data: 0.0002 max mem: 86
Test: [20600/50000] eta: 0:02:04 loss: 0.0362 (0.8077) acc1: 100.0000 (79.8262) acc5: 100.0000 (94.8498) time: 0.0039 data: 0.0002 max mem: 86
Test: [20700/50000] eta: 0:02:04 loss: 0.7428 (0.8134) acc1: 0.0000 (79.7111) acc5: 100.0000 (94.8022) time: 0.0041 data: 0.0002 max mem: 86
Test: [20800/50000] eta: 0:02:03 loss: 1.3721 (0.8242) acc1: 0.0000 (79.5058) acc5: 100.0000 (94.6685) time: 0.0036 data: 0.0002 max mem: 86
Test: [20900/50000] eta: 0:02:03 loss: 0.0012 (0.8234) acc1: 100.0000 (79.5225) acc5: 100.0000 (94.6701) time: 0.0039 data: 0.0002 max mem: 86
Test: [21000/50000] eta: 0:02:02 loss: 0.2224 (0.8257) acc1: 100.0000 (79.4724) acc5: 100.0000 (94.6574) time: 0.0038 data: 0.0002 max mem: 86
Test: [21100/50000] eta: 0:02:02 loss: 0.2699 (0.8258) acc1: 100.0000 (79.4749) acc5: 100.0000 (94.6590) time: 0.0039 data: 0.0002 max mem: 86
Test: [21200/50000] eta: 0:02:02 loss: 0.2430 (0.8297) acc1: 100.0000 (79.3925) acc5: 100.0000 (94.6465) time: 0.0039 data: 0.0002 max mem: 86
Test: [21300/50000] eta: 0:02:01 loss: 0.0005 (0.8302) acc1: 100.0000 (79.3953) acc5: 100.0000 (94.6387) time: 0.0038 data: 0.0002 max mem: 86
Test: [21400/50000] eta: 0:02:01 loss: 0.1168 (0.8315) acc1: 100.0000 (79.3748) acc5: 100.0000 (94.6358) time: 0.0039 data: 0.0002 max mem: 86
Test: [21500/50000] eta: 0:02:00 loss: 0.0134 (0.8331) acc1: 100.0000 (79.3591) acc5: 100.0000 (94.6096) time: 0.0040 data: 0.0002 max mem: 86
Test: [21600/50000] eta: 0:02:00 loss: 0.1121 (0.8323) acc1: 100.0000 (79.3713) acc5: 100.0000 (94.6067) time: 0.0038 data: 0.0002 max mem: 86
Test: [21700/50000] eta: 0:01:59 loss: 0.0110 (0.8337) acc1: 100.0000 (79.3374) acc5: 100.0000 (94.6039) time: 0.0040 data: 0.0003 max mem: 86
Test: [21800/50000] eta: 0:01:59 loss: 0.7444 (0.8395) acc1: 0.0000 (79.1982) acc5: 100.0000 (94.5369) time: 0.0038 data: 0.0002 max mem: 86
Test: [21900/50000] eta: 0:01:58 loss: 0.2301 (0.8426) acc1: 100.0000 (79.1425) acc5: 100.0000 (94.5071) time: 0.0037 data: 0.0002 max mem: 86
Test: [22000/50000] eta: 0:01:58 loss: 0.0000 (0.8459) acc1: 100.0000 (79.0964) acc5: 100.0000 (94.4548) time: 0.0038 data: 0.0002 max mem: 86
Test: [22100/50000] eta: 0:01:57 loss: 0.1825 (0.8474) acc1: 100.0000 (79.0733) acc5: 100.0000 (94.4392) time: 0.0036 data: 0.0002 max mem: 86
Test: [22200/50000] eta: 0:01:57 loss: 0.0235 (0.8493) acc1: 100.0000 (79.0370) acc5: 100.0000 (94.4327) time: 0.0037 data: 0.0002 max mem: 86
Test: [22300/50000] eta: 0:01:56 loss: 1.0514 (0.8507) acc1: 0.0000 (79.0054) acc5: 100.0000 (94.4173) time: 0.0036 data: 0.0002 max mem: 86
Test: [22400/50000] eta: 0:01:56 loss: 0.0888 (0.8552) acc1: 100.0000 (78.9072) acc5: 100.0000 (94.3753) time: 0.0036 data: 0.0002 max mem: 86
Test: [22500/50000] eta: 0:01:55 loss: 0.0557 (0.8541) acc1: 100.0000 (78.9298) acc5: 100.0000 (94.3825) time: 0.0045 data: 0.0003 max mem: 86
Test: [22600/50000] eta: 0:01:55 loss: 0.0050 (0.8534) acc1: 100.0000 (78.9567) acc5: 100.0000 (94.3719) time: 0.0037 data: 0.0002 max mem: 86
Test: [22700/50000] eta: 0:01:54 loss: 0.2502 (0.8564) acc1: 100.0000 (78.8952) acc5: 100.0000 (94.3439) time: 0.0038 data: 0.0002 max mem: 86
Test: [22800/50000] eta: 0:01:54 loss: 0.0013 (0.8606) acc1: 100.0000 (78.7948) acc5: 100.0000 (94.2766) time: 0.0039 data: 0.0002 max mem: 86
Test: [22900/50000] eta: 0:01:53 loss: 0.1173 (0.8620) acc1: 100.0000 (78.7608) acc5: 100.0000 (94.2492) time: 0.0037 data: 0.0002 max mem: 86
Test: [23000/50000] eta: 0:01:53 loss: 0.3043 (0.8637) acc1: 100.0000 (78.7314) acc5: 100.0000 (94.2263) time: 0.0038 data: 0.0002 max mem: 86
Test: [23100/50000] eta: 0:01:52 loss: 1.0789 (0.8663) acc1: 0.0000 (78.6156) acc5: 100.0000 (94.2080) time: 0.0038 data: 0.0002 max mem: 86
Test: [23200/50000] eta: 0:01:52 loss: 0.4022 (0.8728) acc1: 100.0000 (78.5311) acc5: 100.0000 (94.1554) time: 0.0037 data: 0.0002 max mem: 86
Test: [23300/50000] eta: 0:01:52 loss: 1.2780 (0.8800) acc1: 0.0000 (78.4044) acc5: 100.0000 (94.0346) time: 0.0053 data: 0.0013 max mem: 86
Test: [23400/50000] eta: 0:01:51 loss: 0.0102 (0.8784) acc1: 100.0000 (78.4411) acc5: 100.0000 (94.0473) time: 0.0041 data: 0.0002 max mem: 86
Test: [23500/50000] eta: 0:01:51 loss: 1.7649 (0.8824) acc1: 0.0000 (78.3584) acc5: 100.0000 (94.0088) time: 0.0040 data: 0.0002 max mem: 86
Test: [23600/50000] eta: 0:01:50 loss: 0.0051 (0.8853) acc1: 100.0000 (78.3314) acc5: 100.0000 (93.9791) time: 0.0043 data: 0.0002 max mem: 86
Test: [23700/50000] eta: 0:01:50 loss: 0.2883 (0.8852) acc1: 100.0000 (78.3511) acc5: 100.0000 (93.9707) time: 0.0040 data: 0.0002 max mem: 86
Test: [23800/50000] eta: 0:01:49 loss: 0.0004 (0.8854) acc1: 100.0000 (78.3581) acc5: 100.0000 (93.9582) time: 0.0041 data: 0.0002 max mem: 86
Test: [23900/50000] eta: 0:01:49 loss: 0.0361 (0.8858) acc1: 100.0000 (78.3566) acc5: 100.0000 (93.9417) time: 0.0040 data: 0.0002 max mem: 86
Test: [24000/50000] eta: 0:01:49 loss: 1.5626 (0.8937) acc1: 0.0000 (78.2051) acc5: 100.0000 (93.8628) time: 0.0042 data: 0.0002 max mem: 86
Test: [24100/50000] eta: 0:01:48 loss: 0.1333 (0.8958) acc1: 100.0000 (78.1503) acc5: 100.0000 (93.8467) time: 0.0038 data: 0.0002 max mem: 86
Test: [24200/50000] eta: 0:01:48 loss: 0.0076 (0.8986) acc1: 100.0000 (78.0340) acc5: 100.0000 (93.8391) time: 0.0040 data: 0.0002 max mem: 86
Test: [24300/50000] eta: 0:01:47 loss: 1.0008 (0.9013) acc1: 0.0000 (77.9433) acc5: 100.0000 (93.8315) time: 0.0039 data: 0.0002 max mem: 86
Test: [24400/50000] eta: 0:01:47 loss: 0.1565 (0.9017) acc1: 100.0000 (77.9394) acc5: 100.0000 (93.8281) time: 0.0041 data: 0.0002 max mem: 86
Test: [24500/50000] eta: 0:01:46 loss: 0.0557 (0.9064) acc1: 100.0000 (77.8499) acc5: 100.0000 (93.7839) time: 0.0044 data: 0.0003 max mem: 86
Test: [24600/50000] eta: 0:01:46 loss: 0.0539 (0.9073) acc1: 100.0000 (77.8383) acc5: 100.0000 (93.7889) time: 0.0047 data: 0.0002 max mem: 86
Test: [24700/50000] eta: 0:01:46 loss: 2.4288 (0.9127) acc1: 0.0000 (77.7337) acc5: 100.0000 (93.7250) time: 0.0041 data: 0.0002 max mem: 86
Test: [24800/50000] eta: 0:01:45 loss: 0.0067 (0.9131) acc1: 100.0000 (77.7146) acc5: 100.0000 (93.7261) time: 0.0041 data: 0.0002 max mem: 86
Test: [24900/50000] eta: 0:01:45 loss: 0.3004 (0.9131) acc1: 100.0000 (77.7037) acc5: 100.0000 (93.7312) time: 0.0041 data: 0.0002 max mem: 86
Test: [25000/50000] eta: 0:01:44 loss: 1.6132 (0.9186) acc1: 0.0000 (77.5929) acc5: 100.0000 (93.6643) time: 0.0042 data: 0.0002 max mem: 86
Test: [25100/50000] eta: 0:01:44 loss: 1.4236 (0.9222) acc1: 0.0000 (77.5427) acc5: 100.0000 (93.6178) time: 0.0042 data: 0.0002 max mem: 86
Test: [25200/50000] eta: 0:01:44 loss: 0.2022 (0.9257) acc1: 100.0000 (77.4811) acc5: 100.0000 (93.5637) time: 0.0040 data: 0.0002 max mem: 86
Test: [25300/50000] eta: 0:01:43 loss: 0.8040 (0.9300) acc1: 0.0000 (77.3962) acc5: 100.0000 (93.5220) time: 0.0042 data: 0.0002 max mem: 86
Test: [25400/50000] eta: 0:01:43 loss: 0.0320 (0.9325) acc1: 100.0000 (77.3670) acc5: 100.0000 (93.5042) time: 0.0040 data: 0.0002 max mem: 86
Test: [25500/50000] eta: 0:01:42 loss: 0.4818 (0.9350) acc1: 100.0000 (77.3029) acc5: 100.0000 (93.4630) time: 0.0040 data: 0.0002 max mem: 86
Test: [25600/50000] eta: 0:01:42 loss: 0.3171 (0.9337) acc1: 100.0000 (77.3173) acc5: 100.0000 (93.4768) time: 0.0042 data: 0.0002 max mem: 86
Test: [25700/50000] eta: 0:01:41 loss: 0.9818 (0.9333) acc1: 0.0000 (77.2966) acc5: 100.0000 (93.4789) time: 0.0042 data: 0.0002 max mem: 86
Test: [25800/50000] eta: 0:01:41 loss: 0.7947 (0.9360) acc1: 0.0000 (77.2334) acc5: 100.0000 (93.4382) time: 0.0044 data: 0.0002 max mem: 86
Test: [25900/50000] eta: 0:01:41 loss: 0.1149 (0.9387) acc1: 100.0000 (77.1592) acc5: 100.0000 (93.4095) time: 0.0043 data: 0.0002 max mem: 86
Test: [26000/50000] eta: 0:01:40 loss: 0.0282 (0.9402) acc1: 100.0000 (77.1586) acc5: 100.0000 (93.3810) time: 0.0043 data: 0.0002 max mem: 86
Test: [26100/50000] eta: 0:01:40 loss: 0.0062 (0.9403) acc1: 100.0000 (77.1695) acc5: 100.0000 (93.3834) time: 0.0044 data: 0.0002 max mem: 86
Test: [26200/50000] eta: 0:01:39 loss: 0.8185 (0.9423) acc1: 100.0000 (77.1383) acc5: 100.0000 (93.3514) time: 0.0039 data: 0.0002 max mem: 86
Test: [26300/50000] eta: 0:01:39 loss: 0.0799 (0.9432) acc1: 100.0000 (77.0959) acc5: 100.0000 (93.3425) time: 0.0038 data: 0.0002 max mem: 86
Test: [26400/50000] eta: 0:01:39 loss: 1.7410 (0.9467) acc1: 0.0000 (76.9857) acc5: 100.0000 (93.3184) time: 0.0040 data: 0.0002 max mem: 86
Test: [26500/50000] eta: 0:01:38 loss: 0.3317 (0.9479) acc1: 100.0000 (76.9895) acc5: 100.0000 (93.2984) time: 0.0047 data: 0.0002 max mem: 86
Test: [26600/50000] eta: 0:01:38 loss: 0.2205 (0.9518) acc1: 100.0000 (76.9294) acc5: 100.0000 (93.2559) time: 0.0049 data: 0.0002 max mem: 86
Test: [26700/50000] eta: 0:01:37 loss: 0.0376 (0.9516) acc1: 100.0000 (76.9409) acc5: 100.0000 (93.2474) time: 0.0048 data: 0.0002 max mem: 86
Test: [26800/50000] eta: 0:01:37 loss: 0.0017 (0.9525) acc1: 100.0000 (76.9412) acc5: 100.0000 (93.2279) time: 0.0048 data: 0.0002 max mem: 86
Test: [26900/50000] eta: 0:01:37 loss: 0.0244 (0.9534) acc1: 100.0000 (76.9228) acc5: 100.0000 (93.2307) time: 0.0047 data: 0.0002 max mem: 86
Test: [27000/50000] eta: 0:01:36 loss: 0.3235 (0.9550) acc1: 100.0000 (76.8823) acc5: 100.0000 (93.2225) time: 0.0048 data: 0.0002 max mem: 86
Test: [27100/50000] eta: 0:01:36 loss: 1.2845 (0.9566) acc1: 0.0000 (76.8348) acc5: 100.0000 (93.1995) time: 0.0046 data: 0.0002 max mem: 86
Test: [27200/50000] eta: 0:01:36 loss: 0.4835 (0.9589) acc1: 100.0000 (76.7876) acc5: 100.0000 (93.1730) time: 0.0045 data: 0.0002 max mem: 86
Test: [27300/50000] eta: 0:01:35 loss: 0.0237 (0.9588) acc1: 100.0000 (76.7847) acc5: 100.0000 (93.1578) time: 0.0044 data: 0.0002 max mem: 86
Test: [27400/50000] eta: 0:01:35 loss: 0.0135 (0.9583) acc1: 100.0000 (76.8038) acc5: 100.0000 (93.1572) time: 0.0044 data: 0.0002 max mem: 86
Test: [27500/50000] eta: 0:01:34 loss: 1.2116 (0.9596) acc1: 0.0000 (76.7863) acc5: 100.0000 (93.1457) time: 0.0044 data: 0.0002 max mem: 86
Test: [27600/50000] eta: 0:01:34 loss: 0.1187 (0.9596) acc1: 100.0000 (76.7834) acc5: 100.0000 (93.1452) time: 0.0044 data: 0.0002 max mem: 86
Test: [27700/50000] eta: 0:01:34 loss: 0.0066 (0.9588) acc1: 100.0000 (76.7987) acc5: 100.0000 (93.1447) time: 0.0043 data: 0.0002 max mem: 86
Test: [27800/50000] eta: 0:01:33 loss: 0.0399 (0.9574) acc1: 100.0000 (76.8426) acc5: 100.0000 (93.1549) time: 0.0044 data: 0.0002 max mem: 86
Test: [27900/50000] eta: 0:01:33 loss: 0.1268 (0.9612) acc1: 100.0000 (76.8073) acc5: 100.0000 (93.1149) time: 0.0045 data: 0.0002 max mem: 86
Test: [28000/50000] eta: 0:01:32 loss: 0.0496 (0.9647) acc1: 100.0000 (76.7223) acc5: 100.0000 (93.0824) time: 0.0062 data: 0.0002 max mem: 86
Test: [28100/50000] eta: 0:01:32 loss: 0.0062 (0.9628) acc1: 100.0000 (76.7695) acc5: 100.0000 (93.0928) time: 0.0040 data: 0.0002 max mem: 86
Test: [28200/50000] eta: 0:01:31 loss: 0.3135 (0.9621) acc1: 100.0000 (76.7810) acc5: 100.0000 (93.0854) time: 0.0037 data: 0.0002 max mem: 86
Test: [28300/50000] eta: 0:01:31 loss: 0.0000 (0.9602) acc1: 100.0000 (76.8135) acc5: 100.0000 (93.1063) time: 0.0041 data: 0.0002 max mem: 86
Test: [28400/50000] eta: 0:01:31 loss: 2.3275 (0.9628) acc1: 0.0000 (76.7508) acc5: 100.0000 (93.0777) time: 0.0042 data: 0.0002 max mem: 86
Test: [28500/50000] eta: 0:01:30 loss: 0.0125 (0.9610) acc1: 100.0000 (76.7903) acc5: 100.0000 (93.0880) time: 0.0036 data: 0.0002 max mem: 86
Test: [28600/50000] eta: 0:01:30 loss: 0.0271 (0.9603) acc1: 100.0000 (76.7945) acc5: 100.0000 (93.0981) time: 0.0039 data: 0.0002 max mem: 86
Test: [28700/50000] eta: 0:01:29 loss: 0.0012 (0.9598) acc1: 100.0000 (76.8092) acc5: 100.0000 (93.1013) time: 0.0038 data: 0.0002 max mem: 86
Test: [28800/50000] eta: 0:01:29 loss: 0.0002 (0.9593) acc1: 100.0000 (76.8272) acc5: 100.0000 (93.1009) time: 0.0037 data: 0.0002 max mem: 86
Test: [28900/50000] eta: 0:01:28 loss: 0.5748 (0.9582) acc1: 100.0000 (76.8209) acc5: 100.0000 (93.1110) time: 0.0036 data: 0.0002 max mem: 86
Test: [29000/50000] eta: 0:01:28 loss: 0.0352 (0.9589) acc1: 100.0000 (76.7801) acc5: 100.0000 (93.1175) time: 0.0034 data: 0.0002 max mem: 86
Test: [29100/50000] eta: 0:01:27 loss: 0.0741 (0.9591) acc1: 100.0000 (76.7912) acc5: 100.0000 (93.1136) time: 0.0039 data: 0.0002 max mem: 86
Test: [29200/50000] eta: 0:01:27 loss: 0.0006 (0.9596) acc1: 100.0000 (76.7919) acc5: 100.0000 (93.1167) time: 0.0037 data: 0.0002 max mem: 86
Test: [29300/50000] eta: 0:01:26 loss: 2.7847 (0.9667) acc1: 0.0000 (76.6766) acc5: 100.0000 (93.0310) time: 0.0037 data: 0.0002 max mem: 86
Test: [29400/50000] eta: 0:01:26 loss: 0.4571 (0.9681) acc1: 100.0000 (76.6403) acc5: 100.0000 (93.0104) time: 0.0037 data: 0.0002 max mem: 86
Test: [29500/50000] eta: 0:01:26 loss: 0.0102 (0.9711) acc1: 100.0000 (76.5838) acc5: 100.0000 (92.9833) time: 0.0037 data: 0.0002 max mem: 86
Test: [29600/50000] eta: 0:01:25 loss: 0.0034 (0.9734) acc1: 100.0000 (76.5413) acc5: 100.0000 (92.9665) time: 0.0102 data: 0.0067 max mem: 86
Test: [29700/50000] eta: 0:01:25 loss: 0.1565 (0.9735) acc1: 100.0000 (76.5462) acc5: 100.0000 (92.9632) time: 0.0040 data: 0.0003 max mem: 86
Test: [29800/50000] eta: 0:01:24 loss: 0.0976 (0.9718) acc1: 100.0000 (76.5813) acc5: 100.0000 (92.9801) time: 0.0035 data: 0.0002 max mem: 86
Test: [29900/50000] eta: 0:01:24 loss: 0.0910 (0.9741) acc1: 100.0000 (76.5392) acc5: 100.0000 (92.9434) time: 0.0037 data: 0.0002 max mem: 86
Test: [30000/50000] eta: 0:01:23 loss: 0.0262 (0.9762) acc1: 100.0000 (76.5008) acc5: 100.0000 (92.9302) time: 0.0039 data: 0.0002 max mem: 86
Test: [30100/50000] eta: 0:01:23 loss: 0.4759 (0.9820) acc1: 100.0000 (76.4061) acc5: 100.0000 (92.8640) time: 0.0038 data: 0.0002 max mem: 86
Test: [30200/50000] eta: 0:01:23 loss: 0.0131 (0.9812) acc1: 100.0000 (76.4346) acc5: 100.0000 (92.8810) time: 0.0037 data: 0.0002 max mem: 86
Test: [30300/50000] eta: 0:01:22 loss: 0.0078 (0.9805) acc1: 100.0000 (76.4595) acc5: 100.0000 (92.8814) time: 0.0038 data: 0.0002 max mem: 86
Test: [30400/50000] eta: 0:01:22 loss: 0.0000 (0.9786) acc1: 100.0000 (76.5074) acc5: 100.0000 (92.8884) time: 0.0039 data: 0.0002 max mem: 86
Test: [30500/50000] eta: 0:01:21 loss: 0.0187 (0.9795) acc1: 100.0000 (76.4991) acc5: 100.0000 (92.8756) time: 0.0040 data: 0.0002 max mem: 86
Test: [30600/50000] eta: 0:01:21 loss: 0.0008 (0.9788) acc1: 100.0000 (76.5433) acc5: 100.0000 (92.8760) time: 0.0037 data: 0.0002 max mem: 86
Test: [30700/50000] eta: 0:01:20 loss: 0.0136 (0.9775) acc1: 100.0000 (76.5675) acc5: 100.0000 (92.8862) time: 0.0046 data: 0.0010 max mem: 86
Test: [30800/50000] eta: 0:01:20 loss: 0.0479 (0.9777) acc1: 100.0000 (76.5722) acc5: 100.0000 (92.8704) time: 0.0038 data: 0.0002 max mem: 86
Test: [30900/50000] eta: 0:01:19 loss: 0.2425 (0.9775) acc1: 100.0000 (76.5865) acc5: 100.0000 (92.8708) time: 0.0037 data: 0.0002 max mem: 86
Test: [31000/50000] eta: 0:01:19 loss: 0.6480 (0.9819) acc1: 100.0000 (76.4879) acc5: 100.0000 (92.8260) time: 0.0037 data: 0.0002 max mem: 86
Test: [31100/50000] eta: 0:01:19 loss: 0.0027 (0.9835) acc1: 100.0000 (76.4059) acc5: 100.0000 (92.8009) time: 0.0038 data: 0.0002 max mem: 86
Test: [31200/50000] eta: 0:01:18 loss: 2.3043 (0.9897) acc1: 0.0000 (76.3020) acc5: 100.0000 (92.7310) time: 0.0038 data: 0.0002 max mem: 86
Test: [31300/50000] eta: 0:01:18 loss: 0.0125 (0.9893) acc1: 100.0000 (76.3011) acc5: 100.0000 (92.7351) time: 0.0038 data: 0.0002 max mem: 86
Test: [31400/50000] eta: 0:01:17 loss: 0.0191 (0.9905) acc1: 100.0000 (76.2778) acc5: 100.0000 (92.7168) time: 0.0038 data: 0.0002 max mem: 86
Test: [31500/50000] eta: 0:01:17 loss: 0.1016 (0.9900) acc1: 100.0000 (76.2896) acc5: 100.0000 (92.7272) time: 0.0039 data: 0.0002 max mem: 86
Test: [31600/50000] eta: 0:01:16 loss: 0.9199 (0.9900) acc1: 100.0000 (76.2856) acc5: 100.0000 (92.7217) time: 0.0038 data: 0.0002 max mem: 86
Test: [31700/50000] eta: 0:01:16 loss: 5.9094 (0.9981) acc1: 0.0000 (76.1774) acc5: 0.0000 (92.6248) time: 0.0039 data: 0.0003 max mem: 86
Test: [31800/50000] eta: 0:01:16 loss: 0.5320 (0.9992) acc1: 100.0000 (76.1580) acc5: 100.0000 (92.6197) time: 0.0037 data: 0.0002 max mem: 86
Test: [31900/50000] eta: 0:01:15 loss: 0.0417 (0.9993) acc1: 100.0000 (76.1387) acc5: 100.0000 (92.6272) time: 0.0038 data: 0.0002 max mem: 86
Test: [32000/50000] eta: 0:01:15 loss: 1.1720 (1.0015) acc1: 0.0000 (75.9976) acc5: 100.0000 (92.6315) time: 0.0039 data: 0.0002 max mem: 86
Test: [32100/50000] eta: 0:01:14 loss: 0.0168 (1.0000) acc1: 100.0000 (76.0350) acc5: 100.0000 (92.6420) time: 0.0039 data: 0.0002 max mem: 86
Test: [32200/50000] eta: 0:01:14 loss: 1.3763 (1.0011) acc1: 0.0000 (75.9976) acc5: 100.0000 (92.6182) time: 0.0038 data: 0.0002 max mem: 86
Test: [32300/50000] eta: 0:01:13 loss: 0.0001 (1.0009) acc1: 100.0000 (76.0193) acc5: 100.0000 (92.6256) time: 0.0039 data: 0.0002 max mem: 86
Test: [32400/50000] eta: 0:01:13 loss: 0.4792 (1.0012) acc1: 100.0000 (76.0100) acc5: 100.0000 (92.6052) time: 0.0039 data: 0.0002 max mem: 86
Test: [32500/50000] eta: 0:01:12 loss: 0.1603 (1.0021) acc1: 100.0000 (76.0100) acc5: 100.0000 (92.5879) time: 0.0038 data: 0.0002 max mem: 86
Test: [32600/50000] eta: 0:01:12 loss: 0.8276 (1.0068) acc1: 0.0000 (75.9271) acc5: 100.0000 (92.5248) time: 0.0038 data: 0.0002 max mem: 86
Test: [32700/50000] eta: 0:01:12 loss: 0.0924 (1.0085) acc1: 100.0000 (75.9029) acc5: 100.0000 (92.4926) time: 0.0039 data: 0.0002 max mem: 86
Test: [32800/50000] eta: 0:01:11 loss: 1.1209 (1.0095) acc1: 100.0000 (75.8940) acc5: 100.0000 (92.4758) time: 0.0039 data: 0.0002 max mem: 86
Test: [32900/50000] eta: 0:01:11 loss: 0.7599 (1.0134) acc1: 0.0000 (75.7940) acc5: 100.0000 (92.4470) time: 0.0041 data: 0.0002 max mem: 86
Test: [33000/50000] eta: 0:01:10 loss: 0.0403 (1.0133) acc1: 100.0000 (75.8007) acc5: 100.0000 (92.4396) time: 0.0039 data: 0.0002 max mem: 86
Test: [33100/50000] eta: 0:01:10 loss: 0.0076 (1.0135) acc1: 100.0000 (75.8104) acc5: 100.0000 (92.4353) time: 0.0037 data: 0.0002 max mem: 86
Test: [33200/50000] eta: 0:01:09 loss: 0.9662 (1.0160) acc1: 0.0000 (75.7537) acc5: 100.0000 (92.4099) time: 0.0040 data: 0.0002 max mem: 86
Test: [33300/50000] eta: 0:01:09 loss: 0.3771 (1.0177) acc1: 100.0000 (75.6884) acc5: 100.0000 (92.4056) time: 0.0039 data: 0.0002 max mem: 86
Test: [33400/50000] eta: 0:01:09 loss: 1.1264 (1.0185) acc1: 0.0000 (75.6654) acc5: 100.0000 (92.3954) time: 0.0038 data: 0.0002 max mem: 86
Test: [33500/50000] eta: 0:01:08 loss: 0.0021 (1.0160) acc1: 100.0000 (75.7231) acc5: 100.0000 (92.4181) time: 0.0038 data: 0.0002 max mem: 86
Test: [33600/50000] eta: 0:01:08 loss: 0.0493 (1.0149) acc1: 100.0000 (75.7269) acc5: 100.0000 (92.4347) time: 0.0036 data: 0.0002 max mem: 86
Test: [33700/50000] eta: 0:01:07 loss: 1.4782 (1.0162) acc1: 0.0000 (75.7010) acc5: 100.0000 (92.4275) time: 0.0039 data: 0.0002 max mem: 86
Test: [33800/50000] eta: 0:01:07 loss: 2.0012 (1.0179) acc1: 0.0000 (75.6664) acc5: 100.0000 (92.4056) time: 0.0037 data: 0.0002 max mem: 86
Test: [33900/50000] eta: 0:01:06 loss: 0.2956 (1.0205) acc1: 100.0000 (75.6261) acc5: 100.0000 (92.3719) time: 0.0038 data: 0.0002 max mem: 86
Test: [34000/50000] eta: 0:01:06 loss: 0.0472 (1.0215) acc1: 100.0000 (75.6272) acc5: 100.0000 (92.3502) time: 0.0039 data: 0.0002 max mem: 86
Test: [34100/50000] eta: 0:01:06 loss: 1.8225 (1.0236) acc1: 0.0000 (75.5755) acc5: 100.0000 (92.3316) time: 0.0039 data: 0.0002 max mem: 86
Test: [34200/50000] eta: 0:01:05 loss: 0.0572 (1.0243) acc1: 100.0000 (75.5709) acc5: 100.0000 (92.3277) time: 0.0037 data: 0.0002 max mem: 86
Test: [34300/50000] eta: 0:01:05 loss: 0.0000 (1.0229) acc1: 100.0000 (75.6100) acc5: 100.0000 (92.3268) time: 0.0040 data: 0.0002 max mem: 86
Test: [34400/50000] eta: 0:01:04 loss: 0.0172 (1.0238) acc1: 100.0000 (75.5908) acc5: 100.0000 (92.3142) time: 0.0049 data: 0.0002 max mem: 86
Test: [34500/50000] eta: 0:01:04 loss: 2.5449 (1.0254) acc1: 0.0000 (75.5572) acc5: 100.0000 (92.2988) time: 0.0043 data: 0.0002 max mem: 86
Test: [34600/50000] eta: 0:01:03 loss: 0.1552 (1.0262) acc1: 100.0000 (75.5556) acc5: 100.0000 (92.2835) time: 0.0043 data: 0.0002 max mem: 86
Test: [34700/50000] eta: 0:01:03 loss: 0.4390 (1.0281) acc1: 100.0000 (75.4935) acc5: 100.0000 (92.2596) time: 0.0044 data: 0.0002 max mem: 86
Test: [34800/50000] eta: 0:01:03 loss: 0.0076 (1.0276) acc1: 100.0000 (75.5151) acc5: 100.0000 (92.2588) time: 0.0041 data: 0.0002 max mem: 86
Test: [34900/50000] eta: 0:01:02 loss: 0.7986 (1.0291) acc1: 100.0000 (75.4792) acc5: 100.0000 (92.2495) time: 0.0042 data: 0.0002 max mem: 86
Test: [35000/50000] eta: 0:01:02 loss: 0.0039 (1.0286) acc1: 100.0000 (75.4921) acc5: 100.0000 (92.2545) time: 0.0042 data: 0.0002 max mem: 86
Test: [35100/50000] eta: 0:01:01 loss: 0.0001 (1.0286) acc1: 100.0000 (75.5021) acc5: 100.0000 (92.2538) time: 0.0041 data: 0.0002 max mem: 86
Test: [35200/50000] eta: 0:01:01 loss: 0.0196 (1.0284) acc1: 100.0000 (75.4922) acc5: 100.0000 (92.2587) time: 0.0041 data: 0.0002 max mem: 86
Test: [35300/50000] eta: 0:01:01 loss: 0.8688 (1.0297) acc1: 0.0000 (75.4766) acc5: 100.0000 (92.2608) time: 0.0041 data: 0.0002 max mem: 86
Test: [35400/50000] eta: 0:01:00 loss: 0.0061 (1.0310) acc1: 100.0000 (75.4668) acc5: 100.0000 (92.2432) time: 0.0039 data: 0.0002 max mem: 86
Test: [35500/50000] eta: 0:01:00 loss: 0.3687 (1.0314) acc1: 100.0000 (75.4373) acc5: 100.0000 (92.2509) time: 0.0040 data: 0.0002 max mem: 86
Test: [35600/50000] eta: 0:00:59 loss: 0.0745 (1.0329) acc1: 100.0000 (75.3996) acc5: 100.0000 (92.2334) time: 0.0042 data: 0.0002 max mem: 86
Test: [35700/50000] eta: 0:00:59 loss: 0.0009 (1.0327) acc1: 100.0000 (75.4153) acc5: 100.0000 (92.2271) time: 0.0042 data: 0.0002 max mem: 86
Test: [35800/50000] eta: 0:00:58 loss: 0.0018 (1.0322) acc1: 100.0000 (75.4309) acc5: 100.0000 (92.2293) time: 0.0042 data: 0.0002 max mem: 86
Test: [35900/50000] eta: 0:00:58 loss: 0.3446 (1.0322) acc1: 100.0000 (75.4352) acc5: 100.0000 (92.2258) time: 0.0041 data: 0.0002 max mem: 86
Test: [36000/50000] eta: 0:00:58 loss: 0.0002 (1.0324) acc1: 100.0000 (75.4312) acc5: 100.0000 (92.2252) time: 0.0040 data: 0.0002 max mem: 86
Test: [36100/50000] eta: 0:00:57 loss: 0.0874 (1.0323) acc1: 100.0000 (75.4411) acc5: 100.0000 (92.2246) time: 0.0044 data: 0.0002 max mem: 86
Test: [36200/50000] eta: 0:00:57 loss: 0.0008 (1.0320) acc1: 100.0000 (75.4620) acc5: 100.0000 (92.2157) time: 0.0041 data: 0.0002 max mem: 86
Test: [36300/50000] eta: 0:00:56 loss: 1.0876 (1.0326) acc1: 0.0000 (75.4332) acc5: 100.0000 (92.2068) time: 0.0045 data: 0.0002 max mem: 86
Test: [36400/50000] eta: 0:00:56 loss: 0.0012 (1.0310) acc1: 100.0000 (75.4677) acc5: 100.0000 (92.2172) time: 0.0044 data: 0.0002 max mem: 86
Test: [36500/50000] eta: 0:00:56 loss: 0.1963 (1.0367) acc1: 100.0000 (75.3925) acc5: 100.0000 (92.1372) time: 0.0045 data: 0.0002 max mem: 86
Test: [36600/50000] eta: 0:00:55 loss: 0.8977 (1.0400) acc1: 0.0000 (75.3367) acc5: 100.0000 (92.1013) time: 0.0043 data: 0.0002 max mem: 86
Test: [36700/50000] eta: 0:00:55 loss: 1.2821 (1.0421) acc1: 100.0000 (75.3031) acc5: 100.0000 (92.0765) time: 0.0044 data: 0.0002 max mem: 86
Test: [36800/50000] eta: 0:00:54 loss: 0.3187 (1.0422) acc1: 100.0000 (75.2860) acc5: 100.0000 (92.0790) time: 0.0050 data: 0.0002 max mem: 86
Test: [36900/50000] eta: 0:00:54 loss: 0.5414 (1.0421) acc1: 100.0000 (75.2852) acc5: 100.0000 (92.0761) time: 0.0044 data: 0.0002 max mem: 86
Test: [37000/50000] eta: 0:00:54 loss: 0.0000 (1.0410) acc1: 100.0000 (75.3142) acc5: 100.0000 (92.0759) time: 0.0043 data: 0.0002 max mem: 86
Test: [37100/50000] eta: 0:00:53 loss: 0.4119 (1.0430) acc1: 100.0000 (75.2594) acc5: 100.0000 (92.0514) time: 0.0044 data: 0.0002 max mem: 86
Test: [37200/50000] eta: 0:00:53 loss: 0.5268 (1.0443) acc1: 100.0000 (75.2292) acc5: 100.0000 (92.0352) time: 0.0044 data: 0.0002 max mem: 86
Test: [37300/50000] eta: 0:00:52 loss: 0.0056 (1.0465) acc1: 100.0000 (75.1588) acc5: 100.0000 (92.0243) time: 0.0045 data: 0.0002 max mem: 86
Test: [37400/50000] eta: 0:00:52 loss: 1.4897 (1.0469) acc1: 0.0000 (75.1611) acc5: 100.0000 (92.0029) time: 0.0043 data: 0.0002 max mem: 86
Test: [37500/50000] eta: 0:00:52 loss: 0.0313 (1.0487) acc1: 100.0000 (75.1100) acc5: 100.0000 (91.9815) time: 0.0043 data: 0.0002 max mem: 86
Test: [37600/50000] eta: 0:00:51 loss: 0.3074 (1.0492) acc1: 100.0000 (75.0911) acc5: 100.0000 (91.9869) time: 0.0044 data: 0.0002 max mem: 86
Test: [37700/50000] eta: 0:00:51 loss: 0.0172 (1.0496) acc1: 100.0000 (75.0882) acc5: 100.0000 (91.9790) time: 0.0044 data: 0.0002 max mem: 86
Test: [37800/50000] eta: 0:00:50 loss: 0.0022 (1.0492) acc1: 100.0000 (75.0985) acc5: 100.0000 (91.9817) time: 0.0044 data: 0.0002 max mem: 86
Test: [37900/50000] eta: 0:00:50 loss: 0.0170 (1.0489) acc1: 100.0000 (75.1036) acc5: 100.0000 (91.9765) time: 0.0044 data: 0.0002 max mem: 86
Test: [38000/50000] eta: 0:00:49 loss: 0.0966 (1.0499) acc1: 100.0000 (75.0954) acc5: 100.0000 (91.9502) time: 0.0044 data: 0.0002 max mem: 86
Test: [38100/50000] eta: 0:00:49 loss: 0.0094 (1.0497) acc1: 100.0000 (75.0978) acc5: 100.0000 (91.9556) time: 0.0044 data: 0.0002 max mem: 86
Test: [38200/50000] eta: 0:00:49 loss: 0.0218 (1.0517) acc1: 100.0000 (75.0792) acc5: 100.0000 (91.9191) time: 0.0044 data: 0.0002 max mem: 86
Test: [38300/50000] eta: 0:00:48 loss: 0.0913 (1.0534) acc1: 100.0000 (75.0503) acc5: 100.0000 (91.9010) time: 0.0040 data: 0.0002 max mem: 86
Test: [38400/50000] eta: 0:00:48 loss: 3.0212 (1.0541) acc1: 0.0000 (75.0475) acc5: 100.0000 (91.8830) time: 0.0040 data: 0.0002 max mem: 86
Test: [38500/50000] eta: 0:00:47 loss: 0.0590 (1.0545) acc1: 100.0000 (75.0500) acc5: 100.0000 (91.8677) time: 0.0040 data: 0.0002 max mem: 86
Test: [38600/50000] eta: 0:00:47 loss: 0.0100 (1.0544) acc1: 100.0000 (75.0706) acc5: 100.0000 (91.8681) time: 0.0043 data: 0.0001 max mem: 86
Test: [38700/50000] eta: 0:00:47 loss: 0.4904 (1.0564) acc1: 100.0000 (75.0239) acc5: 100.0000 (91.8478) time: 0.0044 data: 0.0002 max mem: 86
Test: [38800/50000] eta: 0:00:46 loss: 0.4960 (1.0580) acc1: 100.0000 (74.9852) acc5: 100.0000 (91.8250) time: 0.0045 data: 0.0002 max mem: 86
Test: [38900/50000] eta: 0:00:46 loss: 0.0769 (1.0580) acc1: 100.0000 (74.9852) acc5: 100.0000 (91.8305) time: 0.0044 data: 0.0002 max mem: 86
Test: [39000/50000] eta: 0:00:45 loss: 0.0051 (1.0587) acc1: 100.0000 (74.9904) acc5: 100.0000 (91.8207) time: 0.0045 data: 0.0002 max mem: 86
Test: [39100/50000] eta: 0:00:45 loss: 0.0018 (1.0566) acc1: 100.0000 (75.0313) acc5: 100.0000 (91.8391) time: 0.0043 data: 0.0002 max mem: 86
Test: [39200/50000] eta: 0:00:45 loss: 0.0012 (1.0582) acc1: 100.0000 (74.9802) acc5: 100.0000 (91.8242) time: 0.0042 data: 0.0002 max mem: 86
Test: [39300/50000] eta: 0:00:44 loss: 0.1853 (1.0605) acc1: 100.0000 (74.9396) acc5: 100.0000 (91.7865) time: 0.0041 data: 0.0002 max mem: 86
Test: [39400/50000] eta: 0:00:44 loss: 0.6406 (1.0621) acc1: 0.0000 (74.9042) acc5: 100.0000 (91.7667) time: 0.0042 data: 0.0002 max mem: 86
Test: [39500/50000] eta: 0:00:43 loss: 0.1936 (1.0617) acc1: 100.0000 (74.9095) acc5: 100.0000 (91.7698) time: 0.0042 data: 0.0002 max mem: 86
Test: [39600/50000] eta: 0:00:43 loss: 0.0425 (1.0628) acc1: 100.0000 (74.8946) acc5: 100.0000 (91.7527) time: 0.0041 data: 0.0002 max mem: 86
Test: [39700/50000] eta: 0:00:42 loss: 0.0153 (1.0645) acc1: 100.0000 (74.8747) acc5: 100.0000 (91.7357) time: 0.0042 data: 0.0002 max mem: 86
Test: [39800/50000] eta: 0:00:42 loss: 0.0769 (1.0651) acc1: 100.0000 (74.8675) acc5: 100.0000 (91.7238) time: 0.0041 data: 0.0002 max mem: 86
Test: [39900/50000] eta: 0:00:42 loss: 0.0277 (1.0658) acc1: 100.0000 (74.8528) acc5: 100.0000 (91.7170) time: 0.0051 data: 0.0002 max mem: 86
Test: [40000/50000] eta: 0:00:41 loss: 0.6600 (1.0687) acc1: 100.0000 (74.8131) acc5: 100.0000 (91.6777) time: 0.0047 data: 0.0002 max mem: 86
Test: [40100/50000] eta: 0:00:41 loss: 0.0090 (1.0668) acc1: 100.0000 (74.8560) acc5: 100.0000 (91.6935) time: 0.0043 data: 0.0002 max mem: 86
Test: [40200/50000] eta: 0:00:40 loss: 0.0264 (1.0652) acc1: 100.0000 (74.8887) acc5: 100.0000 (91.7067) time: 0.0043 data: 0.0002 max mem: 86
Test: [40300/50000] eta: 0:00:40 loss: 0.0093 (1.0654) acc1: 100.0000 (74.8815) acc5: 100.0000 (91.7099) time: 0.0043 data: 0.0002 max mem: 86
Test: [40400/50000] eta: 0:00:40 loss: 0.0218 (1.0666) acc1: 100.0000 (74.8744) acc5: 100.0000 (91.6834) time: 0.0085 data: 0.0045 max mem: 86
Test: [40500/50000] eta: 0:00:39 loss: 1.6256 (1.0685) acc1: 0.0000 (74.8228) acc5: 100.0000 (91.6644) time: 0.0046 data: 0.0002 max mem: 86
Test: [40600/50000] eta: 0:00:39 loss: 0.1739 (1.0713) acc1: 100.0000 (74.7691) acc5: 100.0000 (91.6381) time: 0.0045 data: 0.0002 max mem: 86
Test: [40700/50000] eta: 0:00:38 loss: 2.0720 (1.0730) acc1: 0.0000 (74.7353) acc5: 100.0000 (91.6145) time: 0.0045 data: 0.0002 max mem: 86
Test: [40800/50000] eta: 0:00:38 loss: 0.2095 (1.0723) acc1: 100.0000 (74.7555) acc5: 100.0000 (91.6277) time: 0.0041 data: 0.0002 max mem: 86
Test: [40900/50000] eta: 0:00:38 loss: 0.9667 (1.0716) acc1: 100.0000 (74.7610) acc5: 100.0000 (91.6359) time: 0.0040 data: 0.0002 max mem: 86
Test: [41000/50000] eta: 0:00:37 loss: 0.1690 (1.0745) acc1: 100.0000 (74.7128) acc5: 100.0000 (91.6051) time: 0.0040 data: 0.0002 max mem: 86
Test: [41100/50000] eta: 0:00:37 loss: 0.0553 (1.0729) acc1: 100.0000 (74.7500) acc5: 100.0000 (91.6182) time: 0.0041 data: 0.0002 max mem: 86
Test: [41200/50000] eta: 0:00:36 loss: 0.7887 (1.0736) acc1: 100.0000 (74.7506) acc5: 100.0000 (91.5973) time: 0.0042 data: 0.0002 max mem: 86
Test: [41300/50000] eta: 0:00:36 loss: 0.0577 (1.0742) acc1: 100.0000 (74.7125) acc5: 100.0000 (91.5958) time: 0.0040 data: 0.0002 max mem: 86
Test: [41400/50000] eta: 0:00:35 loss: 0.4766 (1.0766) acc1: 100.0000 (74.6649) acc5: 100.0000 (91.5751) time: 0.0041 data: 0.0002 max mem: 86
Test: [41500/50000] eta: 0:00:35 loss: 0.0224 (1.0772) acc1: 100.0000 (74.6633) acc5: 100.0000 (91.5616) time: 0.0041 data: 0.0002 max mem: 86
Test: [41600/50000] eta: 0:00:35 loss: 0.2120 (1.0775) acc1: 100.0000 (74.6569) acc5: 100.0000 (91.5579) time: 0.0041 data: 0.0002 max mem: 86
Test: [41700/50000] eta: 0:00:34 loss: 0.0137 (1.0766) acc1: 100.0000 (74.6745) acc5: 100.0000 (91.5661) time: 0.0040 data: 0.0001 max mem: 86
Test: [41800/50000] eta: 0:00:34 loss: 0.0237 (1.0781) acc1: 0.0000 (74.6322) acc5: 100.0000 (91.5409) time: 0.0041 data: 0.0002 max mem: 86
Test: [41900/50000] eta: 0:00:33 loss: 2.3601 (1.0831) acc1: 0.0000 (74.5113) acc5: 100.0000 (91.4775) time: 0.0041 data: 0.0002 max mem: 86
Test: [42000/50000] eta: 0:00:33 loss: 0.3241 (1.0853) acc1: 100.0000 (74.4601) acc5: 100.0000 (91.4597) time: 0.0041 data: 0.0002 max mem: 86
Test: [42100/50000] eta: 0:00:33 loss: 0.9063 (1.0870) acc1: 100.0000 (74.4329) acc5: 100.0000 (91.4349) time: 0.0041 data: 0.0002 max mem: 86
Test: [42200/50000] eta: 0:00:32 loss: 0.0135 (1.0873) acc1: 100.0000 (74.4129) acc5: 100.0000 (91.4291) time: 0.0040 data: 0.0002 max mem: 86
Test: [42300/50000] eta: 0:00:32 loss: 0.1794 (1.0892) acc1: 100.0000 (74.3883) acc5: 100.0000 (91.4045) time: 0.0040 data: 0.0002 max mem: 86
Test: [42400/50000] eta: 0:00:31 loss: 0.0941 (1.0898) acc1: 100.0000 (74.3615) acc5: 100.0000 (91.3988) time: 0.0040 data: 0.0002 max mem: 86
Test: [42500/50000] eta: 0:00:31 loss: 0.0792 (1.0902) acc1: 100.0000 (74.3347) acc5: 100.0000 (91.4073) time: 0.0040 data: 0.0002 max mem: 86
Test: [42600/50000] eta: 0:00:30 loss: 0.0957 (1.0900) acc1: 100.0000 (74.3292) acc5: 100.0000 (91.4133) time: 0.0056 data: 0.0002 max mem: 86
Test: [42700/50000] eta: 0:00:30 loss: 0.0044 (1.0891) acc1: 100.0000 (74.3472) acc5: 100.0000 (91.4217) time: 0.0041 data: 0.0002 max mem: 86
Test: [42800/50000] eta: 0:00:30 loss: 0.0020 (1.0896) acc1: 100.0000 (74.3487) acc5: 100.0000 (91.4091) time: 0.0041 data: 0.0002 max mem: 86
Test: [42900/50000] eta: 0:00:29 loss: 0.2166 (1.0903) acc1: 100.0000 (74.3293) acc5: 100.0000 (91.4105) time: 0.0041 data: 0.0002 max mem: 86
Test: [43000/50000] eta: 0:00:29 loss: 0.4386 (1.0907) acc1: 100.0000 (74.3239) acc5: 100.0000 (91.3909) time: 0.0043 data: 0.0002 max mem: 86
Test: [43100/50000] eta: 0:00:28 loss: 0.1530 (1.0923) acc1: 100.0000 (74.2883) acc5: 100.0000 (91.3668) time: 0.0040 data: 0.0002 max mem: 86
Test: [43200/50000] eta: 0:00:28 loss: 0.0065 (1.0916) acc1: 100.0000 (74.3085) acc5: 100.0000 (91.3706) time: 0.0041 data: 0.0002 max mem: 86
Test: [43300/50000] eta: 0:00:27 loss: 0.5499 (1.0922) acc1: 100.0000 (74.2685) acc5: 100.0000 (91.3651) time: 0.0040 data: 0.0002 max mem: 86
Test: [43400/50000] eta: 0:00:27 loss: 0.3370 (1.0916) acc1: 100.0000 (74.2748) acc5: 100.0000 (91.3735) time: 0.0041 data: 0.0002 max mem: 86
Test: [43500/50000] eta: 0:00:27 loss: 0.2863 (1.0939) acc1: 100.0000 (74.2374) acc5: 100.0000 (91.3519) time: 0.0041 data: 0.0002 max mem: 86
Test: [43600/50000] eta: 0:00:26 loss: 0.2812 (1.0930) acc1: 100.0000 (74.2391) acc5: 100.0000 (91.3672) time: 0.0041 data: 0.0002 max mem: 86
Test: [43700/50000] eta: 0:00:26 loss: 0.0013 (1.0925) acc1: 100.0000 (74.2592) acc5: 100.0000 (91.3663) time: 0.0041 data: 0.0002 max mem: 86
Test: [43800/50000] eta: 0:00:25 loss: 0.1880 (1.0916) acc1: 100.0000 (74.2791) acc5: 100.0000 (91.3723) time: 0.0041 data: 0.0002 max mem: 86
Test: [43900/50000] eta: 0:00:25 loss: 0.0064 (1.0920) acc1: 100.0000 (74.2375) acc5: 100.0000 (91.3692) time: 0.0040 data: 0.0002 max mem: 86
Test: [44000/50000] eta: 0:00:25 loss: 0.0775 (1.0924) acc1: 100.0000 (74.2165) acc5: 100.0000 (91.3638) time: 0.0040 data: 0.0002 max mem: 86
Test: [44100/50000] eta: 0:00:24 loss: 0.1228 (1.0920) acc1: 100.0000 (74.2341) acc5: 100.0000 (91.3675) time: 0.0042 data: 0.0002 max mem: 86
Test: [44200/50000] eta: 0:00:24 loss: 1.8513 (1.0932) acc1: 0.0000 (74.2110) acc5: 100.0000 (91.3509) time: 0.0045 data: 0.0002 max mem: 86
Test: [44300/50000] eta: 0:00:23 loss: 3.8736 (1.0965) acc1: 0.0000 (74.1541) acc5: 100.0000 (91.3117) time: 0.0044 data: 0.0002 max mem: 86
Test: [44400/50000] eta: 0:00:23 loss: 0.4045 (1.0968) acc1: 100.0000 (74.1650) acc5: 100.0000 (91.3065) time: 0.0044 data: 0.0002 max mem: 86
Test: [44500/50000] eta: 0:00:22 loss: 0.4718 (1.0967) acc1: 100.0000 (74.1579) acc5: 100.0000 (91.3036) time: 0.0051 data: 0.0002 max mem: 86
Test: [44600/50000] eta: 0:00:22 loss: 0.0673 (1.0961) acc1: 100.0000 (74.1732) acc5: 100.0000 (91.3051) time: 0.0049 data: 0.0002 max mem: 86
Test: [44700/50000] eta: 0:00:22 loss: 1.4266 (1.0967) acc1: 0.0000 (74.1370) acc5: 100.0000 (91.3089) time: 0.0046 data: 0.0002 max mem: 86
Test: [44800/50000] eta: 0:00:21 loss: 0.0431 (1.0960) acc1: 100.0000 (74.1479) acc5: 100.0000 (91.3216) time: 0.0043 data: 0.0002 max mem: 86
Test: [44900/50000] eta: 0:00:21 loss: 0.0027 (1.0956) acc1: 100.0000 (74.1632) acc5: 100.0000 (91.3276) time: 0.0042 data: 0.0002 max mem: 86
Test: [45000/50000] eta: 0:00:20 loss: 2.0405 (1.0988) acc1: 0.0000 (74.0917) acc5: 100.0000 (91.3024) time: 0.0041 data: 0.0002 max mem: 86
Test: [45100/50000] eta: 0:00:20 loss: 0.0499 (1.0986) acc1: 100.0000 (74.1026) acc5: 100.0000 (91.2907) time: 0.0046 data: 0.0002 max mem: 86
Test: [45200/50000] eta: 0:00:20 loss: 0.0301 (1.0984) acc1: 100.0000 (74.1112) acc5: 100.0000 (91.2944) time: 0.0043 data: 0.0002 max mem: 86
Test: [45300/50000] eta: 0:00:19 loss: 0.2390 (1.0996) acc1: 100.0000 (74.0889) acc5: 100.0000 (91.2629) time: 0.0042 data: 0.0002 max mem: 86
Test: [45400/50000] eta: 0:00:19 loss: 1.2460 (1.1034) acc1: 0.0000 (74.0006) acc5: 100.0000 (91.2227) time: 0.0042 data: 0.0002 max mem: 86
Test: [45500/50000] eta: 0:00:18 loss: 0.6810 (1.1048) acc1: 100.0000 (73.9588) acc5: 100.0000 (91.2266) time: 0.0043 data: 0.0002 max mem: 86
Test: [45600/50000] eta: 0:00:18 loss: 0.2011 (1.1071) acc1: 100.0000 (73.9194) acc5: 100.0000 (91.2041) time: 0.0043 data: 0.0002 max mem: 86
Test: [45700/50000] eta: 0:00:17 loss: 0.0228 (1.1068) acc1: 100.0000 (73.9240) acc5: 100.0000 (91.2059) time: 0.0043 data: 0.0002 max mem: 86
Test: [45800/50000] eta: 0:00:17 loss: 0.0003 (1.1058) acc1: 100.0000 (73.9416) acc5: 100.0000 (91.2185) time: 0.0041 data: 0.0002 max mem: 86
Test: [45900/50000] eta: 0:00:17 loss: 0.1972 (1.1048) acc1: 100.0000 (73.9657) acc5: 100.0000 (91.2289) time: 0.0042 data: 0.0002 max mem: 86
Test: [46000/50000] eta: 0:00:16 loss: 0.1110 (1.1049) acc1: 100.0000 (73.9745) acc5: 100.0000 (91.2176) time: 0.0043 data: 0.0002 max mem: 86
Test: [46100/50000] eta: 0:00:16 loss: 0.6403 (1.1047) acc1: 100.0000 (73.9745) acc5: 100.0000 (91.2236) time: 0.0043 data: 0.0002 max mem: 86
Test: [46200/50000] eta: 0:00:15 loss: 1.0495 (1.1057) acc1: 100.0000 (73.9703) acc5: 100.0000 (91.2145) time: 0.0043 data: 0.0002 max mem: 86
Test: [46300/50000] eta: 0:00:15 loss: 0.4804 (1.1053) acc1: 100.0000 (73.9595) acc5: 100.0000 (91.2183) time: 0.0042 data: 0.0002 max mem: 86
Test: [46400/50000] eta: 0:00:15 loss: 0.0027 (1.1038) acc1: 100.0000 (73.9812) acc5: 100.0000 (91.2351) time: 0.0045 data: 0.0002 max mem: 86
Test: [46500/50000] eta: 0:00:14 loss: 0.0935 (1.1043) acc1: 100.0000 (73.9575) acc5: 100.0000 (91.2346) time: 0.0040 data: 0.0002 max mem: 86
Test: [46600/50000] eta: 0:00:14 loss: 0.0400 (1.1054) acc1: 100.0000 (73.9297) acc5: 100.0000 (91.2255) time: 0.0039 data: 0.0002 max mem: 86
Test: [46700/50000] eta: 0:00:13 loss: 0.0019 (1.1049) acc1: 100.0000 (73.9406) acc5: 100.0000 (91.2207) time: 0.0040 data: 0.0002 max mem: 86
Test: [46800/50000] eta: 0:00:13 loss: 0.0435 (1.1051) acc1: 100.0000 (73.9429) acc5: 100.0000 (91.2160) time: 0.0040 data: 0.0002 max mem: 86
Test: [46900/50000] eta: 0:00:12 loss: 0.0004 (1.1039) acc1: 100.0000 (73.9707) acc5: 100.0000 (91.2283) time: 0.0064 data: 0.0026 max mem: 86
Test: [47000/50000] eta: 0:00:12 loss: 0.0368 (1.1030) acc1: 100.0000 (73.9771) acc5: 100.0000 (91.2470) time: 0.0040 data: 0.0002 max mem: 86
Test: [47100/50000] eta: 0:00:12 loss: 0.0096 (1.1024) acc1: 100.0000 (73.9814) acc5: 100.0000 (91.2550) time: 0.0041 data: 0.0002 max mem: 86
Test: [47200/50000] eta: 0:00:11 loss: 0.1077 (1.1020) acc1: 100.0000 (73.9772) acc5: 100.0000 (91.2608) time: 0.0041 data: 0.0002 max mem: 86
Test: [47300/50000] eta: 0:00:11 loss: 0.0041 (1.1010) acc1: 100.0000 (73.9984) acc5: 100.0000 (91.2729) time: 0.0041 data: 0.0002 max mem: 86
Test: [47400/50000] eta: 0:00:10 loss: 0.6002 (1.1012) acc1: 100.0000 (73.9816) acc5: 100.0000 (91.2871) time: 0.0040 data: 0.0002 max mem: 86
Test: [47500/50000] eta: 0:00:10 loss: 0.0593 (1.1010) acc1: 100.0000 (73.9816) acc5: 100.0000 (91.2907) time: 0.0041 data: 0.0002 max mem: 86
Test: [47600/50000] eta: 0:00:10 loss: 0.1147 (1.1010) acc1: 100.0000 (73.9795) acc5: 100.0000 (91.2985) time: 0.0041 data: 0.0002 max mem: 86
Test: [47700/50000] eta: 0:00:09 loss: 0.0013 (1.0998) acc1: 100.0000 (74.0068) acc5: 100.0000 (91.3042) time: 0.0042 data: 0.0002 max mem: 86
Test: [47800/50000] eta: 0:00:09 loss: 0.0000 (1.0988) acc1: 100.0000 (74.0298) acc5: 100.0000 (91.3140) time: 0.0041 data: 0.0002 max mem: 86
Test: [47900/50000] eta: 0:00:08 loss: 0.0004 (1.0977) acc1: 100.0000 (74.0590) acc5: 100.0000 (91.3259) time: 0.0041 data: 0.0002 max mem: 86
Test: [48000/50000] eta: 0:00:08 loss: 0.0048 (1.0966) acc1: 100.0000 (74.0818) acc5: 100.0000 (91.3335) time: 0.0041 data: 0.0002 max mem: 86
Test: [48100/50000] eta: 0:00:07 loss: 1.0288 (1.0997) acc1: 100.0000 (74.0276) acc5: 100.0000 (91.2996) time: 0.0041 data: 0.0002 max mem: 86
Test: [48200/50000] eta: 0:00:07 loss: 0.0181 (1.0999) acc1: 100.0000 (74.0296) acc5: 100.0000 (91.2927) time: 0.0041 data: 0.0002 max mem: 86
Test: [48300/50000] eta: 0:00:07 loss: 0.0095 (1.0998) acc1: 100.0000 (74.0399) acc5: 100.0000 (91.2921) time: 0.0044 data: 0.0002 max mem: 86
Test: [48400/50000] eta: 0:00:06 loss: 0.1439 (1.1010) acc1: 100.0000 (74.0171) acc5: 100.0000 (91.2874) time: 0.0041 data: 0.0002 max mem: 86
Test: [48500/50000] eta: 0:00:06 loss: 2.2339 (1.1049) acc1: 0.0000 (73.9449) acc5: 100.0000 (91.2435) time: 0.0043 data: 0.0002 max mem: 86
Test: [48600/50000] eta: 0:00:05 loss: 0.0188 (1.1055) acc1: 100.0000 (73.9244) acc5: 100.0000 (91.2409) time: 0.0041 data: 0.0002 max mem: 86
Test: [48700/50000] eta: 0:00:05 loss: 0.1100 (1.1054) acc1: 100.0000 (73.9163) acc5: 100.0000 (91.2527) time: 0.0041 data: 0.0002 max mem: 86
Test: [48800/50000] eta: 0:00:05 loss: 1.7498 (1.1053) acc1: 0.0000 (73.9206) acc5: 100.0000 (91.2584) time: 0.0042 data: 0.0002 max mem: 86
Test: [48900/50000] eta: 0:00:04 loss: 0.3690 (1.1059) acc1: 100.0000 (73.8983) acc5: 100.0000 (91.2660) time: 0.0043 data: 0.0002 max mem: 86
Test: [49000/50000] eta: 0:00:04 loss: 0.1133 (1.1072) acc1: 100.0000 (73.8699) acc5: 100.0000 (91.2573) time: 0.0042 data: 0.0002 max mem: 86
Test: [49100/50000] eta: 0:00:03 loss: 0.1942 (1.1067) acc1: 100.0000 (73.8906) acc5: 100.0000 (91.2609) time: 0.0046 data: 0.0003 max mem: 86
Test: [49200/50000] eta: 0:00:03 loss: 0.0112 (1.1064) acc1: 100.0000 (73.9030) acc5: 100.0000 (91.2624) time: 0.0044 data: 0.0002 max mem: 86
Test: [49300/50000] eta: 0:00:02 loss: 0.0015 (1.1046) acc1: 100.0000 (73.9478) acc5: 100.0000 (91.2760) time: 0.0042 data: 0.0001 max mem: 86
Test: [49400/50000] eta: 0:00:02 loss: 0.6341 (1.1039) acc1: 0.0000 (73.9459) acc5: 100.0000 (91.2897) time: 0.0044 data: 0.0002 max mem: 86
Test: [49500/50000] eta: 0:00:02 loss: 0.0015 (1.1020) acc1: 100.0000 (73.9884) acc5: 100.0000 (91.3052) time: 0.0043 data: 0.0002 max mem: 86
Test: [49600/50000] eta: 0:00:01 loss: 0.0005 (1.1002) acc1: 100.0000 (74.0287) acc5: 100.0000 (91.3207) time: 0.0044 data: 0.0002 max mem: 86
Test: [49700/50000] eta: 0:00:01 loss: 0.0000 (1.0989) acc1: 100.0000 (74.0629) acc5: 100.0000 (91.3342) time: 0.0043 data: 0.0002 max mem: 86
Test: [49800/50000] eta: 0:00:00 loss: 0.0000 (1.0973) acc1: 100.0000 (74.1049) acc5: 100.0000 (91.3496) time: 0.0042 data: 0.0002 max mem: 86
Test: [49900/50000] eta: 0:00:00 loss: 0.5711 (1.0966) acc1: 100.0000 (74.1188) acc5: 100.0000 (91.3589) time: 0.0043 data: 0.0002 max mem: 86
Test: Total time: 0:03:30
* Acc@1 74.054 Acc@5 91.340
```
## 应用场景
### 算法类别
图像分类
### 热点行业
制造，能源，交通，网安
### 源码仓库及问题反馈
https://developer.hpccube.com/codes/modelzoo/mobilenet_v3_tvm
### 参考
https://github.com/open-mmlab/mmpretrain
mobilenet.onnx 0 → 100644
View file @ cea8529d
File added
model.pt 0 → 100644
View file @ cea8529d
File added
presets.py 0 → 100644
View file @ cea8529d
from torchvision.transforms import autoaugment, transforms
class ClassificationPresetTrain:
def __init__(self, crop_size, mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225), hflip_prob=0.5,
auto_augment_policy=None, random_erase_prob=0.0):
trans = [transforms.RandomResizedCrop(crop_size)]
if hflip_prob > 0:
trans.append(transforms.RandomHorizontalFlip(hflip_prob))
if auto_augment_policy is not None:
aa_policy = autoaugment.AutoAugmentPolicy(auto_augment_policy)
trans.append(autoaugment.AutoAugment(policy=aa_policy))
trans.extend([
transforms.ToTensor(),
transforms.Normalize(mean=mean, std=std),
])
if random_erase_prob > 0:
trans.append(transforms.RandomErasing(p=random_erase_prob))
self.transforms = transforms.Compose(trans)
def __call__(self, img):
return self.transforms(img)
class ClassificationPresetEval:
def __init__(self, crop_size, resize_size=256, mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225)):
self.transforms = transforms.Compose([
transforms.Resize(resize_size),
transforms.CenterCrop(crop_size),
transforms.ToTensor(),
transforms.Normalize(mean=mean, std=std),
])
def __call__(self, img):
return self.transforms(img)
requirement.txt 0 → 100644
View file @ cea8529d
sphinx==2.4.4
sphinx-gallery>=0.9.0
sphinx-copybutton>=0.3.1
matplotlib
numpy
-e git+git://github.com/pytorch/pytorch_sphinx_theme.git#egg=pytorch_sphinx_theme
run.sh 0 → 100644
View file @ cea8529d
export MIOPEN_DEBUG_CONV_IMPLICIT_GEMM=0
export HIP_VISIBLE_DEVICES=2
python val_onnx.py --test-only --data-path /parastor/DL_DATA/ImageNet-pytorch --model mobilenet_v3_large --b 1 --pretrained
test.py 0 → 100644
View file @ cea8529d
"""Test script for torch module"""
import torch
import time
import tvm
from tvm.contrib.torch import compile
import torch.onnx as onnx
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
x = torch.rand([1, 3, 224, 224]).to(device)
#checkpoint = torch.load("./yolov8n.pt")
#model.load(weights='yolov8n.pt')
#torch.onnx.export(model.predictor.model, x, "./yolov8n.onnx")
#model.export(format='onnx',imgsz=[384, 640], device="cuda")
model_jit = torch.jit.load("./model.pt")
option = {
"input_infos": [
("x", (1, 3, 224, 224)),
],
"default_dtype": "float32",
"export_dir": "pytorch_compiled",
"num_outputs": 1,
"tuning_n_trials": 0, # set zero to skip tuning
"tuning_log_file": "tuning.log",
"target": "rocm --libs=miopen,rocblas",
"device": tvm.rocm(),
}
pytorch_tvm_module = compile(model_jit, option)
print("Run PyTorch...")
for i in range(1):
t = time.time()
# module = torch.jit.load("./model_tvm.pt");
# outputs=module([x])
outputs = pytorch_tvm_module.forward([x])
torch.cuda.synchronize()
print(1000 * (time.time() - t))
print(outputs[0].shape)
torch.result.log 0 → 100644
View file @ cea8529d
[root@dcunode5 mobilenet]# python val_orig.py --test-only --data-path /parastor/DL_DATA/ImageNet-pytorch --model mobilenet_v3_large --b 1 --pretrained
Not using distributed mode
Namespace(apex=False, apex_opt_level='O1', auto_augment=None, batch_size=1, cache_dataset=False, data_path='/parastor/DL_DATA/ImageNet-pytorch', device='cuda', dist_url='env://', distributed=False, epochs=90, lr=0.1, lr_gamma=0.1, lr_step_size=30, model='mobilenet_v3_large', momentum=0.9, opt='sgd', output_dir='.', pretrained=True, print_freq=10, random_erase=0.0, resume='', start_epoch=0, sync_bn=False, test_only=True, weight_decay=0.0001, workers=16, world_size=1)
Loading data
Loading training data
Took 256.0931315422058
Loading validation data
Creating data loaders
Creating model
Test: [ 0/50000] eta: 2 days, 7:14:43 loss: 3.7605 (3.7605) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 3.9777 data: 0.4047 max mem: 29
Test: [ 100/50000] eta: 0:43:41 loss: 0.0005 (0.5272) acc1: 100.0000 (88.1188) acc5: 100.0000 (97.0297) time: 0.0133 data: 0.0002 max mem: 29
Test: [ 200/50000] eta: 0:27:17 loss: 0.1722 (0.6195) acc1: 100.0000 (82.5871) acc5: 100.0000 (96.5174) time: 0.0132 data: 0.0002 max mem: 29
Test: [ 300/50000] eta: 0:21:58 loss: 0.0325 (0.6624) acc1: 100.0000 (82.0598) acc5: 100.0000 (96.0133) time: 0.0146 data: 0.0002 max mem: 29
Test: [ 400/50000] eta: 0:19:24 loss: 0.0283 (0.7113) acc1: 100.0000 (81.7955) acc5: 100.0000 (94.7631) time: 0.0138 data: 0.0002 max mem: 29
Test: [ 500/50000] eta: 0:17:51 loss: 0.0000 (0.6293) acc1: 100.0000 (83.8323) acc5: 100.0000 (95.4092) time: 0.0147 data: 0.0002 max mem: 29
Test: [ 600/50000] eta: 0:16:51 loss: 0.0002 (0.5730) acc1: 100.0000 (85.3577) acc5: 100.0000 (95.8403) time: 0.0145 data: 0.0002 max mem: 29
Test: [ 700/50000] eta: 0:16:08 loss: 0.0023 (0.5215) acc1: 100.0000 (86.5906) acc5: 100.0000 (96.2910) time: 0.0145 data: 0.0002 max mem: 29
Test: [ 800/50000] eta: 0:15:33 loss: 0.0003 (0.4884) acc1: 100.0000 (87.5156) acc5: 100.0000 (96.5044) time: 0.0140 data: 0.0002 max mem: 29
Test: [ 900/50000] eta: 0:15:00 loss: 0.0004 (0.4812) acc1: 100.0000 (87.6804) acc5: 100.0000 (96.8923) time: 0.0138 data: 0.0002 max mem: 29
Test: [ 1000/50000] eta: 0:14:35 loss: 0.0004 (0.4654) acc1: 100.0000 (88.4116) acc5: 100.0000 (97.0030) time: 0.0135 data: 0.0003 max mem: 29
Test: [ 1100/50000] eta: 0:14:12 loss: 0.0158 (0.5022) acc1: 100.0000 (87.4659) acc5: 100.0000 (96.5486) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 1200/50000] eta: 0:13:55 loss: 0.0051 (0.4699) acc1: 100.0000 (88.1765) acc5: 100.0000 (96.8360) time: 0.0135 data: 0.0002 max mem: 29
Test: [ 1300/50000] eta: 0:13:43 loss: 0.0049 (0.4499) acc1: 100.0000 (88.8547) acc5: 100.0000 (96.9254) time: 0.0151 data: 0.0002 max mem: 29
Test: [ 1400/50000] eta: 0:13:29 loss: 0.0190 (0.5285) acc1: 100.0000 (87.2234) acc5: 100.0000 (96.4311) time: 0.0132 data: 0.0002 max mem: 29
Test: [ 1500/50000] eta: 0:13:16 loss: 0.0029 (0.5403) acc1: 100.0000 (87.1419) acc5: 100.0000 (96.1359) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 1600/50000] eta: 0:13:04 loss: 0.0291 (0.5636) acc1: 100.0000 (86.8832) acc5: 100.0000 (96.0025) time: 0.0138 data: 0.0002 max mem: 29
Test: [ 1700/50000] eta: 0:12:56 loss: 0.0790 (0.6317) acc1: 100.0000 (85.3028) acc5: 100.0000 (95.7084) time: 0.0143 data: 0.0002 max mem: 29
Test: [ 1800/50000] eta: 0:12:47 loss: 0.8301 (0.6861) acc1: 0.0000 (83.8978) acc5: 100.0000 (95.2804) time: 0.0128 data: 0.0002 max mem: 29
Test: [ 1900/50000] eta: 0:12:39 loss: 0.1593 (0.6977) acc1: 100.0000 (83.2194) acc5: 100.0000 (95.4235) time: 0.0134 data: 0.0002 max mem: 29
Test: [ 2000/50000] eta: 0:12:32 loss: 0.0165 (0.7040) acc1: 100.0000 (83.2084) acc5: 100.0000 (95.4023) time: 0.0147 data: 0.0002 max mem: 29
Test: [ 2100/50000] eta: 0:12:25 loss: 0.0156 (0.7113) acc1: 100.0000 (82.8653) acc5: 100.0000 (95.4307) time: 0.0132 data: 0.0002 max mem: 29
Test: [ 2200/50000] eta: 0:12:20 loss: 0.0035 (0.7085) acc1: 100.0000 (82.9623) acc5: 100.0000 (95.4112) time: 0.0138 data: 0.0002 max mem: 29
Test: [ 2300/50000] eta: 0:12:14 loss: 0.0000 (0.7477) acc1: 100.0000 (82.3990) acc5: 100.0000 (95.0456) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 2400/50000] eta: 0:12:07 loss: 0.0144 (0.7674) acc1: 100.0000 (81.7576) acc5: 100.0000 (94.9188) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 2500/50000] eta: 0:12:03 loss: 0.1349 (0.7727) acc1: 100.0000 (81.7673) acc5: 100.0000 (94.7621) time: 0.0143 data: 0.0002 max mem: 29
Test: [ 2600/50000] eta: 0:11:59 loss: 0.0001 (0.7679) acc1: 100.0000 (81.8916) acc5: 100.0000 (94.6943) time: 0.0139 data: 0.0002 max mem: 29
Test: [ 2700/50000] eta: 0:11:54 loss: 0.2303 (0.7688) acc1: 100.0000 (81.7845) acc5: 100.0000 (94.6686) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 2800/50000] eta: 0:11:51 loss: 0.1836 (0.7950) acc1: 100.0000 (81.1853) acc5: 100.0000 (94.5734) time: 0.0146 data: 0.0002 max mem: 29
Test: [ 2900/50000] eta: 0:11:49 loss: 0.0785 (0.7940) acc1: 100.0000 (81.0410) acc5: 100.0000 (94.5881) time: 0.0144 data: 0.0002 max mem: 29
Test: [ 3000/50000] eta: 0:11:45 loss: 0.8794 (0.8190) acc1: 100.0000 (80.5398) acc5: 100.0000 (94.3019) time: 0.0128 data: 0.0002 max mem: 29
Test: [ 3100/50000] eta: 0:11:40 loss: 0.1602 (0.8417) acc1: 100.0000 (79.8130) acc5: 100.0000 (94.3567) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 3200/50000] eta: 0:11:36 loss: 0.0120 (0.8620) acc1: 100.0000 (79.2565) acc5: 100.0000 (94.0331) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 3300/50000] eta: 0:11:33 loss: 0.1979 (0.8830) acc1: 100.0000 (78.8549) acc5: 100.0000 (93.9109) time: 0.0135 data: 0.0002 max mem: 29
Test: [ 3400/50000] eta: 0:11:29 loss: 0.8700 (0.8987) acc1: 0.0000 (78.3593) acc5: 100.0000 (93.7959) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 3500/50000] eta: 0:11:25 loss: 0.0001 (0.9023) acc1: 100.0000 (77.8920) acc5: 100.0000 (93.7446) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 3600/50000] eta: 0:11:21 loss: 0.0002 (0.8921) acc1: 100.0000 (78.0894) acc5: 100.0000 (93.7795) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 3700/50000] eta: 0:11:18 loss: 1.0290 (0.8984) acc1: 0.0000 (77.5736) acc5: 100.0000 (93.8665) time: 0.0137 data: 0.0002 max mem: 29
Test: [ 3800/50000] eta: 0:11:15 loss: 0.0186 (0.9037) acc1: 100.0000 (77.2691) acc5: 100.0000 (93.9227) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 3900/50000] eta: 0:11:12 loss: 0.0721 (0.9016) acc1: 100.0000 (77.3135) acc5: 100.0000 (93.9503) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 4000/50000] eta: 0:11:09 loss: 0.0351 (0.9084) acc1: 100.0000 (77.3057) acc5: 100.0000 (93.8765) time: 0.0136 data: 0.0002 max mem: 29
Test: [ 4100/50000] eta: 0:11:06 loss: 0.0029 (0.8997) acc1: 100.0000 (77.4689) acc5: 100.0000 (94.0015) time: 0.0137 data: 0.0002 max mem: 29
Test: [ 4200/50000] eta: 0:11:04 loss: 0.0013 (0.8858) acc1: 100.0000 (77.7910) acc5: 100.0000 (94.1204) time: 0.0150 data: 0.0002 max mem: 29
Test: [ 4300/50000] eta: 0:11:02 loss: 0.0011 (0.8741) acc1: 100.0000 (78.1214) acc5: 100.0000 (94.1641) time: 0.0133 data: 0.0002 max mem: 29
Test: [ 4400/50000] eta: 0:10:59 loss: 0.0000 (0.8666) acc1: 100.0000 (78.1641) acc5: 100.0000 (94.2513) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 4500/50000] eta: 0:10:58 loss: 0.0001 (0.8528) acc1: 100.0000 (78.5603) acc5: 100.0000 (94.3568) time: 0.0144 data: 0.0002 max mem: 29
Test: [ 4600/50000] eta: 0:10:56 loss: 0.0002 (0.8414) acc1: 100.0000 (78.8524) acc5: 100.0000 (94.3708) time: 0.0148 data: 0.0002 max mem: 29
Test: [ 4700/50000] eta: 0:10:55 loss: 0.0031 (0.8295) acc1: 100.0000 (79.1534) acc5: 100.0000 (94.4905) time: 0.0149 data: 0.0002 max mem: 29
Test: [ 4800/50000] eta: 0:10:53 loss: 0.0007 (0.8202) acc1: 100.0000 (79.4210) acc5: 100.0000 (94.5220) time: 0.0141 data: 0.0002 max mem: 29
Test: [ 4900/50000] eta: 0:10:50 loss: 0.0011 (0.8136) acc1: 100.0000 (79.6164) acc5: 100.0000 (94.5725) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 5000/50000] eta: 0:10:49 loss: 0.0025 (0.8080) acc1: 100.0000 (79.7041) acc5: 100.0000 (94.6411) time: 0.0144 data: 0.0002 max mem: 29
Test: [ 5100/50000] eta: 0:10:47 loss: 0.3220 (0.8063) acc1: 100.0000 (79.7099) acc5: 100.0000 (94.6481) time: 0.0144 data: 0.0002 max mem: 29
Test: [ 5200/50000] eta: 0:10:45 loss: 0.0062 (0.8012) acc1: 100.0000 (79.8500) acc5: 100.0000 (94.6164) time: 0.0143 data: 0.0002 max mem: 29
Test: [ 5300/50000] eta: 0:10:43 loss: 0.0002 (0.7945) acc1: 100.0000 (80.0415) acc5: 100.0000 (94.6237) time: 0.0145 data: 0.0002 max mem: 29
Test: [ 5400/50000] eta: 0:10:41 loss: 0.0015 (0.7994) acc1: 100.0000 (80.1889) acc5: 100.0000 (94.5751) time: 0.0145 data: 0.0002 max mem: 29
Test: [ 5500/50000] eta: 0:10:40 loss: 0.0039 (0.8008) acc1: 100.0000 (80.2218) acc5: 100.0000 (94.5646) time: 0.0137 data: 0.0002 max mem: 29
Test: [ 5600/50000] eta: 0:10:37 loss: 0.0009 (0.7973) acc1: 100.0000 (80.3428) acc5: 100.0000 (94.5903) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 5700/50000] eta: 0:10:35 loss: 0.0045 (0.8016) acc1: 100.0000 (80.3368) acc5: 100.0000 (94.4922) time: 0.0132 data: 0.0002 max mem: 29
Test: [ 5800/50000] eta: 0:10:33 loss: 0.0441 (0.8010) acc1: 100.0000 (80.4172) acc5: 100.0000 (94.4665) time: 0.0127 data: 0.0002 max mem: 29
Test: [ 5900/50000] eta: 0:10:30 loss: 0.0016 (0.8022) acc1: 100.0000 (80.5287) acc5: 100.0000 (94.4416) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 6000/50000] eta: 0:10:28 loss: 0.0303 (0.8104) acc1: 100.0000 (80.4199) acc5: 100.0000 (94.4176) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 6100/50000] eta: 0:10:26 loss: 0.0044 (0.8116) acc1: 100.0000 (80.3475) acc5: 100.0000 (94.4271) time: 0.0144 data: 0.0002 max mem: 29
Test: [ 6200/50000] eta: 0:10:24 loss: 0.0818 (0.8107) acc1: 100.0000 (80.3580) acc5: 100.0000 (94.3719) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 6300/50000] eta: 0:10:22 loss: 0.0063 (0.8139) acc1: 100.0000 (80.2888) acc5: 100.0000 (94.3977) time: 0.0137 data: 0.0002 max mem: 29
Test: [ 6400/50000] eta: 0:10:20 loss: 0.0055 (0.8195) acc1: 100.0000 (80.3000) acc5: 100.0000 (94.2978) time: 0.0146 data: 0.0002 max mem: 29
Test: [ 6500/50000] eta: 0:10:18 loss: 0.0001 (0.8117) acc1: 100.0000 (80.4645) acc5: 100.0000 (94.3393) time: 0.0140 data: 0.0002 max mem: 29
Test: [ 6600/50000] eta: 0:10:17 loss: 0.0092 (0.8027) acc1: 100.0000 (80.6696) acc5: 100.0000 (94.4251) time: 0.0145 data: 0.0002 max mem: 29
Test: [ 6700/50000] eta: 0:10:15 loss: 0.0012 (0.7978) acc1: 100.0000 (80.7790) acc5: 100.0000 (94.4337) time: 0.0140 data: 0.0002 max mem: 29
Test: [ 6800/50000] eta: 0:10:13 loss: 0.0000 (0.7927) acc1: 100.0000 (80.8558) acc5: 100.0000 (94.4861) time: 0.0132 data: 0.0002 max mem: 29
Test: [ 6900/50000] eta: 0:10:11 loss: 0.0122 (0.7864) acc1: 100.0000 (81.0172) acc5: 100.0000 (94.5225) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 7000/50000] eta: 0:10:09 loss: 0.0003 (0.7786) acc1: 100.0000 (81.2170) acc5: 100.0000 (94.5579) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 7100/50000] eta: 0:10:07 loss: 0.0009 (0.7743) acc1: 100.0000 (81.3407) acc5: 100.0000 (94.5782) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 7200/50000] eta: 0:10:05 loss: 0.0001 (0.7686) acc1: 100.0000 (81.4331) acc5: 100.0000 (94.6257) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 7300/50000] eta: 0:10:03 loss: 0.0002 (0.7626) acc1: 100.0000 (81.6053) acc5: 100.0000 (94.6583) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 7400/50000] eta: 0:10:01 loss: 0.0007 (0.7567) acc1: 100.0000 (81.7457) acc5: 100.0000 (94.7034) time: 0.0131 data: 0.0002 max mem: 29
Test: [ 7500/50000] eta: 0:09:59 loss: 0.0025 (0.7521) acc1: 100.0000 (81.8558) acc5: 100.0000 (94.7074) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 7600/50000] eta: 0:09:57 loss: 0.0618 (0.7581) acc1: 100.0000 (81.7392) acc5: 100.0000 (94.6718) time: 0.0132 data: 0.0002 max mem: 29
Test: [ 7700/50000] eta: 0:09:55 loss: 0.2603 (0.7584) acc1: 100.0000 (81.7167) acc5: 100.0000 (94.6760) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 7800/50000] eta: 0:09:53 loss: 0.3902 (0.7611) acc1: 100.0000 (81.6177) acc5: 100.0000 (94.6545) time: 0.0128 data: 0.0002 max mem: 29
Test: [ 7900/50000] eta: 0:09:51 loss: 0.0121 (0.7564) acc1: 100.0000 (81.7745) acc5: 100.0000 (94.6969) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 8000/50000] eta: 0:09:49 loss: 0.6661 (0.7616) acc1: 0.0000 (81.5523) acc5: 100.0000 (94.6882) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 8100/50000] eta: 0:09:47 loss: 0.0326 (0.7601) acc1: 100.0000 (81.5578) acc5: 100.0000 (94.6797) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 8200/50000] eta: 0:09:45 loss: 0.6533 (0.7703) acc1: 100.0000 (81.4413) acc5: 100.0000 (94.5982) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 8300/50000] eta: 0:09:43 loss: 0.5622 (0.7768) acc1: 100.0000 (81.3637) acc5: 100.0000 (94.5428) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 8400/50000] eta: 0:09:41 loss: 0.9667 (0.7885) acc1: 0.0000 (80.9666) acc5: 100.0000 (94.4768) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 8500/50000] eta: 0:09:39 loss: 0.0256 (0.7869) acc1: 100.0000 (81.0022) acc5: 100.0000 (94.5065) time: 0.0130 data: 0.0002 max mem: 29
Test: [ 8600/50000] eta: 0:09:37 loss: 0.0389 (0.7950) acc1: 100.0000 (80.8394) acc5: 100.0000 (94.4774) time: 0.0129 data: 0.0002 max mem: 29
Test: [ 8700/50000] eta: 0:09:35 loss: 0.2748 (0.7973) acc1: 100.0000 (80.7378) acc5: 100.0000 (94.4604) time: 0.0128 data: 0.0002 max mem: 29
Test: [ 8800/50000] eta: 0:09:33 loss: 0.1653 (0.8007) acc1: 100.0000 (80.7181) acc5: 100.0000 (94.4438) time: 0.0126 data: 0.0002 max mem: 29
Test: [ 8900/50000] eta: 0:09:32 loss: 0.1470 (0.7986) acc1: 100.0000 (80.7550) acc5: 100.0000 (94.4388) time: 0.0145 data: 0.0002 max mem: 29
Test: [ 9000/50000] eta: 0:09:31 loss: 0.2561 (0.7993) acc1: 100.0000 (80.7133) acc5: 100.0000 (94.4340) time: 0.0142 data: 0.0002 max mem: 29
Test: [ 9100/50000] eta: 0:09:30 loss: 0.0316 (0.7989) acc1: 100.0000 (80.6725) acc5: 100.0000 (94.4621) time: 0.0142 data: 0.0002 max mem: 29
Test: [ 9200/50000] eta: 0:09:28 loss: 0.3239 (0.7978) acc1: 100.0000 (80.6543) acc5: 100.0000 (94.4680) time: 0.0146 data: 0.0002 max mem: 29
Test: [ 9300/50000] eta: 0:09:27 loss: 0.1094 (0.8015) acc1: 100.0000 (80.5075) acc5: 100.0000 (94.4737) time: 0.0148 data: 0.0002 max mem: 29
Test: [ 9400/50000] eta: 0:09:26 loss: 0.2506 (0.8040) acc1: 100.0000 (80.4489) acc5: 100.0000 (94.4580) time: 0.0148 data: 0.0002 max mem: 29
Test: [ 9500/50000] eta: 0:09:24 loss: 0.1578 (0.8035) acc1: 100.0000 (80.4021) acc5: 100.0000 (94.4743) time: 0.0128 data: 0.0002 max mem: 29
Test: [ 9600/50000] eta: 0:09:22 loss: 0.0188 (0.8056) acc1: 100.0000 (80.4187) acc5: 100.0000 (94.4381) time: 0.0127 data: 0.0002 max mem: 29
Test: [ 9700/50000] eta: 0:09:20 loss: 0.4514 (0.8102) acc1: 100.0000 (80.2495) acc5: 100.0000 (94.4233) time: 0.0128 data: 0.0002 max mem: 29
Test: [ 9800/50000] eta: 0:09:18 loss: 0.0612 (0.8095) acc1: 100.0000 (80.2775) acc5: 100.0000 (94.4189) time: 0.0128 data: 0.0002 max mem: 29
Test: [ 9900/50000] eta: 0:09:16 loss: 0.3089 (0.8120) acc1: 100.0000 (80.2242) acc5: 100.0000 (94.4248) time: 0.0127 data: 0.0002 max mem: 29
Test: [10000/50000] eta: 0:09:15 loss: 0.0415 (0.8116) acc1: 100.0000 (80.2320) acc5: 100.0000 (94.4606) time: 0.0127 data: 0.0002 max mem: 29
Test: [10100/50000] eta: 0:09:13 loss: 0.5171 (0.8145) acc1: 100.0000 (80.0614) acc5: 100.0000 (94.4758) time: 0.0128 data: 0.0002 max mem: 29
Test: [10200/50000] eta: 0:09:11 loss: 0.0181 (0.8114) acc1: 100.0000 (80.0902) acc5: 100.0000 (94.5103) time: 0.0127 data: 0.0002 max mem: 29
Test: [10300/50000] eta: 0:09:09 loss: 0.0997 (0.8119) acc1: 100.0000 (80.0408) acc5: 100.0000 (94.5345) time: 0.0131 data: 0.0002 max mem: 29
Test: [10400/50000] eta: 0:09:08 loss: 0.0290 (0.8098) acc1: 100.0000 (80.0788) acc5: 100.0000 (94.5582) time: 0.0131 data: 0.0002 max mem: 29
Test: [10500/50000] eta: 0:09:06 loss: 0.0330 (0.8097) acc1: 100.0000 (80.1257) acc5: 100.0000 (94.5624) time: 0.0127 data: 0.0002 max mem: 29
Test: [10600/50000] eta: 0:09:04 loss: 0.1191 (0.8095) acc1: 100.0000 (80.1057) acc5: 100.0000 (94.5477) time: 0.0131 data: 0.0002 max mem: 29
Test: [10700/50000] eta: 0:09:03 loss: 0.0099 (0.8100) acc1: 100.0000 (80.0953) acc5: 100.0000 (94.5426) time: 0.0129 data: 0.0002 max mem: 29
Test: [10800/50000] eta: 0:09:01 loss: 0.2072 (0.8087) acc1: 100.0000 (80.1315) acc5: 100.0000 (94.5468) time: 0.0128 data: 0.0002 max mem: 29
Test: [10900/50000] eta: 0:08:59 loss: 0.0128 (0.8042) acc1: 100.0000 (80.2312) acc5: 100.0000 (94.5785) time: 0.0135 data: 0.0002 max mem: 29
Test: [11000/50000] eta: 0:08:58 loss: 0.0486 (0.8027) acc1: 100.0000 (80.2382) acc5: 100.0000 (94.6187) time: 0.0130 data: 0.0002 max mem: 29
Test: [11100/50000] eta: 0:08:56 loss: 0.0621 (0.8054) acc1: 100.0000 (80.1910) acc5: 100.0000 (94.6041) time: 0.0145 data: 0.0002 max mem: 29
Test: [11200/50000] eta: 0:08:54 loss: 0.0130 (0.8056) acc1: 100.0000 (80.1893) acc5: 100.0000 (94.5719) time: 0.0128 data: 0.0002 max mem: 29
Test: [11300/50000] eta: 0:08:53 loss: 0.0220 (0.8028) acc1: 100.0000 (80.2407) acc5: 100.0000 (94.5934) time: 0.0127 data: 0.0002 max mem: 29
Test: [11400/50000] eta: 0:08:51 loss: 0.0704 (0.8124) acc1: 100.0000 (80.1333) acc5: 100.0000 (94.5180) time: 0.0130 data: 0.0002 max mem: 29
Test: [11500/50000] eta: 0:08:49 loss: 0.0229 (0.8090) acc1: 100.0000 (80.2278) acc5: 100.0000 (94.5135) time: 0.0131 data: 0.0002 max mem: 29
Test: [11600/50000] eta: 0:08:48 loss: 0.3336 (0.8100) acc1: 100.0000 (80.1396) acc5: 100.0000 (94.5005) time: 0.0132 data: 0.0002 max mem: 29
Test: [11700/50000] eta: 0:08:46 loss: 0.0793 (0.8110) acc1: 100.0000 (80.1214) acc5: 100.0000 (94.4877) time: 0.0129 data: 0.0002 max mem: 29
Test: [11800/50000] eta: 0:08:45 loss: 0.0249 (0.8095) acc1: 100.0000 (80.1542) acc5: 100.0000 (94.5174) time: 0.0147 data: 0.0002 max mem: 29
Test: [11900/50000] eta: 0:08:43 loss: 0.0822 (0.8103) acc1: 100.0000 (80.1361) acc5: 100.0000 (94.4711) time: 0.0148 data: 0.0002 max mem: 29
Test: [12000/50000] eta: 0:08:42 loss: 0.0385 (0.8100) acc1: 100.0000 (80.1350) acc5: 100.0000 (94.5088) time: 0.0147 data: 0.0002 max mem: 29
Test: [12100/50000] eta: 0:08:41 loss: 0.2712 (0.8136) acc1: 100.0000 (79.9273) acc5: 100.0000 (94.5376) time: 0.0146 data: 0.0002 max mem: 29
Test: [12200/50000] eta: 0:08:40 loss: 0.0601 (0.8135) acc1: 100.0000 (79.8951) acc5: 100.0000 (94.5496) time: 0.0126 data: 0.0002 max mem: 29
Test: [12300/50000] eta: 0:08:38 loss: 0.0080 (0.8108) acc1: 100.0000 (79.9610) acc5: 100.0000 (94.5695) time: 0.0125 data: 0.0002 max mem: 29
Test: [12400/50000] eta: 0:08:37 loss: 0.0194 (0.8125) acc1: 100.0000 (79.9129) acc5: 100.0000 (94.5488) time: 0.0125 data: 0.0002 max mem: 29
Test: [12500/50000] eta: 0:08:35 loss: 0.2914 (0.8163) acc1: 100.0000 (79.7696) acc5: 100.0000 (94.5444) time: 0.0125 data: 0.0002 max mem: 29
Test: [12600/50000] eta: 0:08:33 loss: 0.0031 (0.8145) acc1: 100.0000 (79.7794) acc5: 100.0000 (94.5877) time: 0.0131 data: 0.0002 max mem: 29
Test: [12700/50000] eta: 0:08:32 loss: 0.0760 (0.8128) acc1: 100.0000 (79.7969) acc5: 100.0000 (94.6146) time: 0.0130 data: 0.0002 max mem: 29
Test: [12800/50000] eta: 0:08:30 loss: 0.0001 (0.8072) acc1: 100.0000 (79.9234) acc5: 100.0000 (94.6489) time: 0.0141 data: 0.0002 max mem: 29
Test: [12900/50000] eta: 0:08:29 loss: 0.0351 (0.8071) acc1: 100.0000 (79.9008) acc5: 100.0000 (94.6516) time: 0.0130 data: 0.0002 max mem: 29
Test: [13000/50000] eta: 0:08:27 loss: 0.0030 (0.8019) acc1: 100.0000 (80.0323) acc5: 100.0000 (94.6927) time: 0.0129 data: 0.0002 max mem: 29
Test: [13100/50000] eta: 0:08:26 loss: 0.0005 (0.7990) acc1: 100.0000 (80.1084) acc5: 100.0000 (94.7180) time: 0.0130 data: 0.0002 max mem: 29
Test: [13200/50000] eta: 0:08:24 loss: 0.0728 (0.7972) acc1: 100.0000 (80.1454) acc5: 100.0000 (94.7352) time: 0.0130 data: 0.0002 max mem: 29
Test: [13300/50000] eta: 0:08:23 loss: 0.6036 (0.7994) acc1: 100.0000 (80.0241) acc5: 100.0000 (94.7372) time: 0.0131 data: 0.0002 max mem: 29
Test: [13400/50000] eta: 0:08:21 loss: 0.0842 (0.8006) acc1: 100.0000 (79.9269) acc5: 100.0000 (94.7392) time: 0.0141 data: 0.0002 max mem: 29
Test: [13500/50000] eta: 0:08:20 loss: 0.3301 (0.7995) acc1: 100.0000 (79.9348) acc5: 100.0000 (94.7634) time: 0.0130 data: 0.0002 max mem: 29
Test: [13600/50000] eta: 0:08:18 loss: 0.4816 (0.8014) acc1: 100.0000 (79.8765) acc5: 100.0000 (94.7651) time: 0.0130 data: 0.0002 max mem: 29
Test: [13700/50000] eta: 0:08:16 loss: 0.0488 (0.8076) acc1: 100.0000 (79.8044) acc5: 100.0000 (94.7376) time: 0.0129 data: 0.0002 max mem: 29
Test: [13800/50000] eta: 0:08:15 loss: 0.0009 (0.8044) acc1: 100.0000 (79.9145) acc5: 100.0000 (94.7685) time: 0.0130 data: 0.0002 max mem: 29
Test: [13900/50000] eta: 0:08:13 loss: 0.2700 (0.8054) acc1: 100.0000 (79.8576) acc5: 100.0000 (94.7917) time: 0.0129 data: 0.0002 max mem: 29
Test: [14000/50000] eta: 0:08:12 loss: 0.0046 (0.8038) acc1: 100.0000 (79.8657) acc5: 100.0000 (94.8218) time: 0.0132 data: 0.0002 max mem: 29
Test: [14100/50000] eta: 0:08:10 loss: 0.6475 (0.8075) acc1: 100.0000 (79.7603) acc5: 100.0000 (94.8089) time: 0.0131 data: 0.0002 max mem: 29
Test: [14200/50000] eta: 0:08:09 loss: 0.0036 (0.8107) acc1: 100.0000 (79.5930) acc5: 100.0000 (94.8032) time: 0.0130 data: 0.0002 max mem: 29
Test: [14300/50000] eta: 0:08:07 loss: 0.8770 (0.8150) acc1: 0.0000 (79.5189) acc5: 100.0000 (94.7416) time: 0.0129 data: 0.0002 max mem: 29
Test: [14400/50000] eta: 0:08:06 loss: 0.0047 (0.8161) acc1: 100.0000 (79.5500) acc5: 100.0000 (94.7365) time: 0.0131 data: 0.0002 max mem: 29
Test: [14500/50000] eta: 0:08:04 loss: 0.0571 (0.8135) acc1: 100.0000 (79.5945) acc5: 100.0000 (94.7659) time: 0.0130 data: 0.0002 max mem: 29
Test: [14600/50000] eta: 0:08:03 loss: 0.0004 (0.8125) acc1: 100.0000 (79.6384) acc5: 100.0000 (94.7743) time: 0.0130 data: 0.0002 max mem: 29
Test: [14700/50000] eta: 0:08:01 loss: 0.0237 (0.8089) acc1: 100.0000 (79.7293) acc5: 100.0000 (94.8031) time: 0.0130 data: 0.0002 max mem: 29
Test: [14800/50000] eta: 0:08:00 loss: 0.0363 (0.8052) acc1: 100.0000 (79.7784) acc5: 100.0000 (94.8382) time: 0.0129 data: 0.0002 max mem: 29
Test: [14900/50000] eta: 0:07:58 loss: 0.0038 (0.8080) acc1: 100.0000 (79.7933) acc5: 100.0000 (94.8259) time: 0.0130 data: 0.0002 max mem: 29
Test: [15000/50000] eta: 0:07:57 loss: 0.1103 (0.8090) acc1: 100.0000 (79.7680) acc5: 100.0000 (94.8203) time: 0.0131 data: 0.0002 max mem: 29
Test: [15100/50000] eta: 0:07:55 loss: 0.0379 (0.8061) acc1: 100.0000 (79.8027) acc5: 100.0000 (94.8414) time: 0.0130 data: 0.0002 max mem: 29
Test: [15200/50000] eta: 0:07:54 loss: 0.2943 (0.8078) acc1: 100.0000 (79.7448) acc5: 100.0000 (94.8622) time: 0.0130 data: 0.0002 max mem: 29
Test: [15300/50000] eta: 0:07:52 loss: 0.0289 (0.8078) acc1: 100.0000 (79.7137) acc5: 100.0000 (94.8827) time: 0.0130 data: 0.0002 max mem: 29
Test: [15400/50000] eta: 0:07:51 loss: 0.0153 (0.8058) acc1: 100.0000 (79.7546) acc5: 100.0000 (94.8964) time: 0.0130 data: 0.0002 max mem: 29
Test: [15500/50000] eta: 0:07:49 loss: 0.0358 (0.8057) acc1: 100.0000 (79.7239) acc5: 100.0000 (94.8907) time: 0.0130 data: 0.0002 max mem: 29
Test: [15600/50000] eta: 0:07:48 loss: 0.1787 (0.8104) acc1: 100.0000 (79.6103) acc5: 100.0000 (94.8529) time: 0.0130 data: 0.0002 max mem: 29
Test: [15700/50000] eta: 0:07:46 loss: 0.2585 (0.8120) acc1: 100.0000 (79.5172) acc5: 100.0000 (94.8602) time: 0.0130 data: 0.0002 max mem: 29
Test: [15800/50000] eta: 0:07:45 loss: 0.0671 (0.8148) acc1: 100.0000 (79.5140) acc5: 100.0000 (94.8168) time: 0.0128 data: 0.0002 max mem: 29
Test: [15900/50000] eta: 0:07:43 loss: 0.0272 (0.8127) acc1: 100.0000 (79.5736) acc5: 100.0000 (94.8368) time: 0.0130 data: 0.0002 max mem: 29
Test: [16000/50000] eta: 0:07:42 loss: 0.0547 (0.8122) acc1: 100.0000 (79.6013) acc5: 100.0000 (94.8503) time: 0.0130 data: 0.0002 max mem: 29
Test: [16100/50000] eta: 0:07:40 loss: 0.0000 (0.8089) acc1: 100.0000 (79.6907) acc5: 100.0000 (94.8699) time: 0.0133 data: 0.0002 max mem: 29
Test: [16200/50000] eta: 0:07:39 loss: 0.0006 (0.8054) acc1: 100.0000 (79.7852) acc5: 100.0000 (94.8954) time: 0.0144 data: 0.0002 max mem: 29
Test: [16300/50000] eta: 0:07:38 loss: 0.0003 (0.8019) acc1: 100.0000 (79.8724) acc5: 100.0000 (94.9267) time: 0.0129 data: 0.0002 max mem: 29
Test: [16400/50000] eta: 0:07:36 loss: 0.0062 (0.8004) acc1: 100.0000 (79.9281) acc5: 100.0000 (94.9332) time: 0.0129 data: 0.0002 max mem: 29
Test: [16500/50000] eta: 0:07:35 loss: 0.0020 (0.8020) acc1: 100.0000 (79.9224) acc5: 100.0000 (94.8973) time: 0.0129 data: 0.0002 max mem: 29
Test: [16600/50000] eta: 0:07:33 loss: 0.0321 (0.8014) acc1: 100.0000 (79.9108) acc5: 100.0000 (94.9160) time: 0.0128 data: 0.0002 max mem: 29
Test: [16700/50000] eta: 0:07:32 loss: 0.0001 (0.7972) acc1: 100.0000 (80.0132) acc5: 100.0000 (94.9464) time: 0.0128 data: 0.0002 max mem: 29
Test: [16800/50000] eta: 0:07:30 loss: 0.0031 (0.7957) acc1: 100.0000 (80.0488) acc5: 100.0000 (94.9646) time: 0.0129 data: 0.0002 max mem: 29
Test: [16900/50000] eta: 0:07:29 loss: 0.0138 (0.7970) acc1: 100.0000 (80.0781) acc5: 100.0000 (94.9707) time: 0.0129 data: 0.0002 max mem: 29
Test: [17000/50000] eta: 0:07:27 loss: 0.0021 (0.7951) acc1: 100.0000 (80.1306) acc5: 100.0000 (94.9768) time: 0.0150 data: 0.0002 max mem: 29
Test: [17100/50000] eta: 0:07:26 loss: 1.9055 (0.7958) acc1: 0.0000 (80.1181) acc5: 100.0000 (94.9594) time: 0.0129 data: 0.0002 max mem: 29
Test: [17200/50000] eta: 0:07:24 loss: 0.0138 (0.7940) acc1: 100.0000 (80.1523) acc5: 100.0000 (94.9770) time: 0.0128 data: 0.0002 max mem: 29
Test: [17300/50000] eta: 0:07:23 loss: 0.4505 (0.7939) acc1: 100.0000 (80.1225) acc5: 100.0000 (94.9829) time: 0.0129 data: 0.0002 max mem: 29
Test: [17400/50000] eta: 0:07:21 loss: 0.0002 (0.7931) acc1: 100.0000 (80.1448) acc5: 100.0000 (94.9601) time: 0.0129 data: 0.0002 max mem: 29
Test: [17500/50000] eta: 0:07:20 loss: 0.3345 (0.7926) acc1: 100.0000 (80.0640) acc5: 100.0000 (94.9774) time: 0.0130 data: 0.0002 max mem: 29
Test: [17600/50000] eta: 0:07:18 loss: 0.0260 (0.7892) acc1: 100.0000 (80.1432) acc5: 100.0000 (95.0003) time: 0.0129 data: 0.0002 max mem: 29
Test: [17700/50000] eta: 0:07:17 loss: 0.0452 (0.7887) acc1: 100.0000 (80.1198) acc5: 100.0000 (95.0172) time: 0.0129 data: 0.0002 max mem: 29
Test: [17800/50000] eta: 0:07:15 loss: 0.0026 (0.7865) acc1: 100.0000 (80.1865) acc5: 100.0000 (95.0171) time: 0.0147 data: 0.0002 max mem: 29
Test: [17900/50000] eta: 0:07:14 loss: 0.0605 (0.7897) acc1: 100.0000 (80.1017) acc5: 100.0000 (95.0226) time: 0.0130 data: 0.0002 max mem: 29
Test: [18000/50000] eta: 0:07:13 loss: 0.1980 (0.7939) acc1: 100.0000 (79.9956) acc5: 100.0000 (95.0114) time: 0.0145 data: 0.0002 max mem: 29
Test: [18100/50000] eta: 0:07:11 loss: 0.0005 (0.7933) acc1: 100.0000 (79.9845) acc5: 100.0000 (95.0113) time: 0.0131 data: 0.0002 max mem: 29
Test: [18200/50000] eta: 0:07:10 loss: 0.0000 (0.7923) acc1: 100.0000 (80.0231) acc5: 100.0000 (95.0113) time: 0.0130 data: 0.0002 max mem: 29
Test: [18300/50000] eta: 0:07:09 loss: 0.0105 (0.7899) acc1: 100.0000 (80.1049) acc5: 100.0000 (95.0221) time: 0.0130 data: 0.0002 max mem: 29
Test: [18400/50000] eta: 0:07:07 loss: 0.0868 (0.7898) acc1: 100.0000 (80.1261) acc5: 100.0000 (95.0166) time: 0.0131 data: 0.0002 max mem: 29
Test: [18500/50000] eta: 0:07:06 loss: 0.0747 (0.7911) acc1: 100.0000 (80.1146) acc5: 100.0000 (94.9949) time: 0.0130 data: 0.0002 max mem: 29
Test: [18600/50000] eta: 0:07:04 loss: 0.0352 (0.7937) acc1: 100.0000 (80.1086) acc5: 100.0000 (94.9734) time: 0.0131 data: 0.0002 max mem: 29
Test: [18700/50000] eta: 0:07:03 loss: 0.0732 (0.7940) acc1: 100.0000 (80.1294) acc5: 100.0000 (94.9575) time: 0.0131 data: 0.0002 max mem: 29
Test: [18800/50000] eta: 0:07:01 loss: 0.0474 (0.7955) acc1: 100.0000 (80.1074) acc5: 100.0000 (94.9418) time: 0.0131 data: 0.0002 max mem: 29
Test: [18900/50000] eta: 0:07:00 loss: 0.1086 (0.7941) acc1: 100.0000 (80.1386) acc5: 100.0000 (94.9474) time: 0.0130 data: 0.0002 max mem: 29
Test: [19000/50000] eta: 0:06:58 loss: 0.0733 (0.7935) acc1: 100.0000 (80.1484) acc5: 100.0000 (94.9529) time: 0.0130 data: 0.0002 max mem: 29
Test: [19100/50000] eta: 0:06:57 loss: 1.5252 (0.7983) acc1: 0.0000 (80.0063) acc5: 100.0000 (94.9322) time: 0.0130 data: 0.0002 max mem: 29
Test: [19200/50000] eta: 0:06:56 loss: 0.0863 (0.8004) acc1: 100.0000 (79.9385) acc5: 100.0000 (94.9273) time: 0.0131 data: 0.0002 max mem: 29
Test: [19300/50000] eta: 0:06:54 loss: 0.7411 (0.7994) acc1: 100.0000 (79.9285) acc5: 100.0000 (94.9277) time: 0.0129 data: 0.0002 max mem: 29
Test: [19400/50000] eta: 0:06:53 loss: 0.0001 (0.8005) acc1: 100.0000 (79.9083) acc5: 100.0000 (94.9281) time: 0.0129 data: 0.0001 max mem: 29
Test: [19500/50000] eta: 0:06:51 loss: 0.0033 (0.8001) acc1: 100.0000 (79.9190) acc5: 100.0000 (94.9182) time: 0.0129 data: 0.0002 max mem: 29
Test: [19600/50000] eta: 0:06:50 loss: 0.0073 (0.8024) acc1: 100.0000 (79.8786) acc5: 100.0000 (94.9033) time: 0.0128 data: 0.0002 max mem: 29
Test: [19700/50000] eta: 0:06:48 loss: 0.0179 (0.8002) acc1: 100.0000 (79.9452) acc5: 100.0000 (94.9140) time: 0.0128 data: 0.0002 max mem: 29
Test: [19800/50000] eta: 0:06:47 loss: 0.1315 (0.8002) acc1: 100.0000 (79.9303) acc5: 100.0000 (94.9043) time: 0.0131 data: 0.0002 max mem: 29
Test: [19900/50000] eta: 0:06:46 loss: 0.0081 (0.7988) acc1: 100.0000 (79.9910) acc5: 100.0000 (94.9148) time: 0.0129 data: 0.0002 max mem: 29
Test: [20000/50000] eta: 0:06:44 loss: 0.1578 (0.7995) acc1: 100.0000 (79.9760) acc5: 100.0000 (94.9053) time: 0.0129 data: 0.0002 max mem: 29
Test: [20100/50000] eta: 0:06:43 loss: 0.1097 (0.8019) acc1: 100.0000 (79.9015) acc5: 100.0000 (94.8858) time: 0.0127 data: 0.0002 max mem: 29
Test: [20200/50000] eta: 0:06:41 loss: 0.0078 (0.8048) acc1: 100.0000 (79.8673) acc5: 100.0000 (94.8567) time: 0.0130 data: 0.0002 max mem: 29
Test: [20300/50000] eta: 0:06:40 loss: 0.0131 (0.8054) acc1: 100.0000 (79.8778) acc5: 100.0000 (94.8278) time: 0.0128 data: 0.0002 max mem: 29
Test: [20400/50000] eta: 0:06:38 loss: 0.0012 (0.8049) acc1: 100.0000 (79.8882) acc5: 100.0000 (94.8434) time: 0.0128 data: 0.0002 max mem: 29
Test: [20500/50000] eta: 0:06:37 loss: 0.3294 (0.8079) acc1: 100.0000 (79.8303) acc5: 100.0000 (94.8490) time: 0.0129 data: 0.0002 max mem: 29
Test: [20600/50000] eta: 0:06:35 loss: 0.0362 (0.8077) acc1: 100.0000 (79.8262) acc5: 100.0000 (94.8498) time: 0.0130 data: 0.0002 max mem: 29
Test: [20700/50000] eta: 0:06:34 loss: 0.7428 (0.8134) acc1: 0.0000 (79.7111) acc5: 100.0000 (94.8022) time: 0.0130 data: 0.0002 max mem: 29
Test: [20800/50000] eta: 0:06:33 loss: 1.3721 (0.8242) acc1: 0.0000 (79.5058) acc5: 100.0000 (94.6685) time: 0.0132 data: 0.0002 max mem: 29
Test: [20900/50000] eta: 0:06:31 loss: 0.0012 (0.8234) acc1: 100.0000 (79.5225) acc5: 100.0000 (94.6701) time: 0.0129 data: 0.0002 max mem: 29
Test: [21000/50000] eta: 0:06:30 loss: 0.2224 (0.8257) acc1: 100.0000 (79.4724) acc5: 100.0000 (94.6574) time: 0.0128 data: 0.0002 max mem: 29
Test: [21100/50000] eta: 0:06:28 loss: 0.2699 (0.8258) acc1: 100.0000 (79.4749) acc5: 100.0000 (94.6590) time: 0.0131 data: 0.0002 max mem: 29
Test: [21200/50000] eta: 0:06:27 loss: 0.2430 (0.8297) acc1: 100.0000 (79.3925) acc5: 100.0000 (94.6465) time: 0.0129 data: 0.0002 max mem: 29
Test: [21300/50000] eta: 0:06:26 loss: 0.0005 (0.8302) acc1: 100.0000 (79.3953) acc5: 100.0000 (94.6387) time: 0.0131 data: 0.0002 max mem: 29
Test: [21400/50000] eta: 0:06:24 loss: 0.1168 (0.8315) acc1: 100.0000 (79.3748) acc5: 100.0000 (94.6358) time: 0.0128 data: 0.0002 max mem: 29
Test: [21500/50000] eta: 0:06:23 loss: 0.0134 (0.8331) acc1: 100.0000 (79.3591) acc5: 100.0000 (94.6096) time: 0.0126 data: 0.0002 max mem: 29
Test: [21600/50000] eta: 0:06:21 loss: 0.1121 (0.8323) acc1: 100.0000 (79.3713) acc5: 100.0000 (94.6067) time: 0.0127 data: 0.0002 max mem: 29
Test: [21700/50000] eta: 0:06:20 loss: 0.0110 (0.8337) acc1: 100.0000 (79.3374) acc5: 100.0000 (94.6039) time: 0.0126 data: 0.0002 max mem: 29
Test: [21800/50000] eta: 0:06:18 loss: 0.7444 (0.8395) acc1: 0.0000 (79.1982) acc5: 100.0000 (94.5369) time: 0.0127 data: 0.0002 max mem: 29
Test: [21900/50000] eta: 0:06:17 loss: 0.2301 (0.8426) acc1: 100.0000 (79.1425) acc5: 100.0000 (94.5071) time: 0.0126 data: 0.0002 max mem: 29
Test: [22000/50000] eta: 0:06:15 loss: 0.0000 (0.8459) acc1: 100.0000 (79.0964) acc5: 100.0000 (94.4548) time: 0.0127 data: 0.0002 max mem: 29
Test: [22100/50000] eta: 0:06:14 loss: 0.1825 (0.8474) acc1: 100.0000 (79.0733) acc5: 100.0000 (94.4392) time: 0.0126 data: 0.0002 max mem: 29
Test: [22200/50000] eta: 0:06:13 loss: 0.0235 (0.8493) acc1: 100.0000 (79.0370) acc5: 100.0000 (94.4327) time: 0.0126 data: 0.0002 max mem: 29
Test: [22300/50000] eta: 0:06:11 loss: 1.0514 (0.8507) acc1: 0.0000 (79.0054) acc5: 100.0000 (94.4173) time: 0.0127 data: 0.0002 max mem: 29
Test: [22400/50000] eta: 0:06:10 loss: 0.0888 (0.8552) acc1: 100.0000 (78.9072) acc5: 100.0000 (94.3753) time: 0.0126 data: 0.0002 max mem: 29
Test: [22500/50000] eta: 0:06:08 loss: 0.0557 (0.8541) acc1: 100.0000 (78.9298) acc5: 100.0000 (94.3825) time: 0.0142 data: 0.0002 max mem: 29
Test: [22600/50000] eta: 0:06:07 loss: 0.0050 (0.8534) acc1: 100.0000 (78.9567) acc5: 100.0000 (94.3719) time: 0.0138 data: 0.0002 max mem: 29
Test: [22700/50000] eta: 0:06:06 loss: 0.2502 (0.8564) acc1: 100.0000 (78.8952) acc5: 100.0000 (94.3439) time: 0.0130 data: 0.0002 max mem: 29
Test: [22800/50000] eta: 0:06:04 loss: 0.0013 (0.8606) acc1: 100.0000 (78.7948) acc5: 100.0000 (94.2766) time: 0.0127 data: 0.0002 max mem: 29
Test: [22900/50000] eta: 0:06:03 loss: 0.1173 (0.8620) acc1: 100.0000 (78.7608) acc5: 100.0000 (94.2492) time: 0.0142 data: 0.0002 max mem: 29
Test: [23000/50000] eta: 0:06:02 loss: 0.3043 (0.8637) acc1: 100.0000 (78.7314) acc5: 100.0000 (94.2263) time: 0.0133 data: 0.0002 max mem: 29
Test: [23100/50000] eta: 0:06:00 loss: 1.0789 (0.8663) acc1: 0.0000 (78.6156) acc5: 100.0000 (94.2080) time: 0.0128 data: 0.0002 max mem: 29
Test: [23200/50000] eta: 0:05:59 loss: 0.4022 (0.8728) acc1: 100.0000 (78.5311) acc5: 100.0000 (94.1554) time: 0.0128 data: 0.0002 max mem: 29
Test: [23300/50000] eta: 0:05:57 loss: 1.2780 (0.8800) acc1: 0.0000 (78.4044) acc5: 100.0000 (94.0346) time: 0.0129 data: 0.0002 max mem: 29
Test: [23400/50000] eta: 0:05:56 loss: 0.0102 (0.8784) acc1: 100.0000 (78.4411) acc5: 100.0000 (94.0473) time: 0.0145 data: 0.0002 max mem: 29
Test: [23500/50000] eta: 0:05:55 loss: 1.7649 (0.8824) acc1: 0.0000 (78.3584) acc5: 100.0000 (94.0088) time: 0.0143 data: 0.0002 max mem: 29
Test: [23600/50000] eta: 0:05:54 loss: 0.0051 (0.8853) acc1: 100.0000 (78.3314) acc5: 100.0000 (93.9791) time: 0.0127 data: 0.0002 max mem: 29
Test: [23700/50000] eta: 0:05:52 loss: 0.2883 (0.8852) acc1: 100.0000 (78.3511) acc5: 100.0000 (93.9707) time: 0.0126 data: 0.0002 max mem: 29
Test: [23800/50000] eta: 0:05:51 loss: 0.0004 (0.8854) acc1: 100.0000 (78.3581) acc5: 100.0000 (93.9582) time: 0.0127 data: 0.0002 max mem: 29
Test: [23900/50000] eta: 0:05:49 loss: 0.0361 (0.8858) acc1: 100.0000 (78.3566) acc5: 100.0000 (93.9417) time: 0.0127 data: 0.0002 max mem: 29
Test: [24000/50000] eta: 0:05:48 loss: 1.5626 (0.8937) acc1: 0.0000 (78.2051) acc5: 100.0000 (93.8628) time: 0.0127 data: 0.0002 max mem: 29
Test: [24100/50000] eta: 0:05:46 loss: 0.1333 (0.8958) acc1: 100.0000 (78.1503) acc5: 100.0000 (93.8467) time: 0.0128 data: 0.0002 max mem: 29
Test: [24200/50000] eta: 0:05:45 loss: 0.0076 (0.8986) acc1: 100.0000 (78.0340) acc5: 100.0000 (93.8391) time: 0.0127 data: 0.0001 max mem: 29
Test: [24300/50000] eta: 0:05:44 loss: 1.0008 (0.9013) acc1: 0.0000 (77.9433) acc5: 100.0000 (93.8315) time: 0.0127 data: 0.0002 max mem: 29
Test: [24400/50000] eta: 0:05:42 loss: 0.1565 (0.9017) acc1: 100.0000 (77.9394) acc5: 100.0000 (93.8281) time: 0.0127 data: 0.0002 max mem: 29
Test: [24500/50000] eta: 0:05:41 loss: 0.0557 (0.9064) acc1: 100.0000 (77.8499) acc5: 100.0000 (93.7839) time: 0.0128 data: 0.0002 max mem: 29
Test: [24600/50000] eta: 0:05:39 loss: 0.0539 (0.9073) acc1: 100.0000 (77.8383) acc5: 100.0000 (93.7889) time: 0.0126 data: 0.0002 max mem: 29
Test: [24700/50000] eta: 0:05:38 loss: 2.4288 (0.9127) acc1: 0.0000 (77.7337) acc5: 100.0000 (93.7250) time: 0.0128 data: 0.0002 max mem: 29
Test: [24800/50000] eta: 0:05:37 loss: 0.0067 (0.9131) acc1: 100.0000 (77.7146) acc5: 100.0000 (93.7261) time: 0.0128 data: 0.0002 max mem: 29
Test: [24900/50000] eta: 0:05:35 loss: 0.3004 (0.9131) acc1: 100.0000 (77.7037) acc5: 100.0000 (93.7312) time: 0.0127 data: 0.0002 max mem: 29
Test: [25000/50000] eta: 0:05:34 loss: 1.6132 (0.9186) acc1: 0.0000 (77.5929) acc5: 100.0000 (93.6643) time: 0.0129 data: 0.0002 max mem: 29
Test: [25100/50000] eta: 0:05:32 loss: 1.4236 (0.9222) acc1: 0.0000 (77.5427) acc5: 100.0000 (93.6178) time: 0.0123 data: 0.0002 max mem: 29
Test: [25200/50000] eta: 0:05:31 loss: 0.2022 (0.9257) acc1: 100.0000 (77.4811) acc5: 100.0000 (93.5637) time: 0.0124 data: 0.0001 max mem: 29
Test: [25300/50000] eta: 0:05:29 loss: 0.8040 (0.9300) acc1: 0.0000 (77.3962) acc5: 100.0000 (93.5220) time: 0.0124 data: 0.0002 max mem: 29
Test: [25400/50000] eta: 0:05:28 loss: 0.0320 (0.9325) acc1: 100.0000 (77.3670) acc5: 100.0000 (93.5042) time: 0.0126 data: 0.0002 max mem: 29
Test: [25500/50000] eta: 0:05:27 loss: 0.4818 (0.9350) acc1: 100.0000 (77.3029) acc5: 100.0000 (93.4630) time: 0.0123 data: 0.0002 max mem: 29
Test: [25600/50000] eta: 0:05:25 loss: 0.3170 (0.9337) acc1: 100.0000 (77.3173) acc5: 100.0000 (93.4768) time: 0.0124 data: 0.0001 max mem: 29
Test: [25700/50000] eta: 0:05:24 loss: 0.9818 (0.9333) acc1: 0.0000 (77.2966) acc5: 100.0000 (93.4789) time: 0.0124 data: 0.0002 max mem: 29
Test: [25800/50000] eta: 0:05:22 loss: 0.7947 (0.9360) acc1: 0.0000 (77.2334) acc5: 100.0000 (93.4382) time: 0.0125 data: 0.0002 max mem: 29
Test: [25900/50000] eta: 0:05:21 loss: 0.1149 (0.9387) acc1: 100.0000 (77.1592) acc5: 100.0000 (93.4095) time: 0.0122 data: 0.0002 max mem: 29
Test: [26000/50000] eta: 0:05:19 loss: 0.0282 (0.9402) acc1: 100.0000 (77.1586) acc5: 100.0000 (93.3810) time: 0.0121 data: 0.0002 max mem: 29
Test: [26100/50000] eta: 0:05:18 loss: 0.0062 (0.9403) acc1: 100.0000 (77.1695) acc5: 100.0000 (93.3834) time: 0.0123 data: 0.0002 max mem: 29
Test: [26200/50000] eta: 0:05:17 loss: 0.8185 (0.9423) acc1: 100.0000 (77.1383) acc5: 100.0000 (93.3514) time: 0.0122 data: 0.0002 max mem: 29
Test: [26300/50000] eta: 0:05:15 loss: 0.0799 (0.9432) acc1: 100.0000 (77.0959) acc5: 100.0000 (93.3425) time: 0.0122 data: 0.0002 max mem: 29
Test: [26400/50000] eta: 0:05:14 loss: 1.7410 (0.9467) acc1: 0.0000 (76.9857) acc5: 100.0000 (93.3184) time: 0.0121 data: 0.0002 max mem: 29
Test: [26500/50000] eta: 0:05:12 loss: 0.3317 (0.9479) acc1: 100.0000 (76.9895) acc5: 100.0000 (93.2984) time: 0.0120 data: 0.0002 max mem: 29
Test: [26600/50000] eta: 0:05:11 loss: 0.2205 (0.9518) acc1: 100.0000 (76.9294) acc5: 100.0000 (93.2559) time: 0.0122 data: 0.0002 max mem: 29
Test: [26700/50000] eta: 0:05:09 loss: 0.0376 (0.9516) acc1: 100.0000 (76.9409) acc5: 100.0000 (93.2474) time: 0.0121 data: 0.0002 max mem: 29
Test: [26800/50000] eta: 0:05:08 loss: 0.0017 (0.9525) acc1: 100.0000 (76.9412) acc5: 100.0000 (93.2279) time: 0.0121 data: 0.0002 max mem: 29
Test: [26900/50000] eta: 0:05:07 loss: 0.0244 (0.9534) acc1: 100.0000 (76.9228) acc5: 100.0000 (93.2307) time: 0.0122 data: 0.0002 max mem: 29
Test: [27000/50000] eta: 0:05:05 loss: 0.3235 (0.9550) acc1: 100.0000 (76.8823) acc5: 100.0000 (93.2225) time: 0.0121 data: 0.0001 max mem: 29
Test: [27100/50000] eta: 0:05:04 loss: 1.2845 (0.9566) acc1: 0.0000 (76.8348) acc5: 100.0000 (93.1995) time: 0.0128 data: 0.0002 max mem: 29
Test: [27200/50000] eta: 0:05:02 loss: 0.4835 (0.9589) acc1: 100.0000 (76.7876) acc5: 100.0000 (93.1730) time: 0.0128 data: 0.0002 max mem: 29
Test: [27300/50000] eta: 0:05:01 loss: 0.0237 (0.9588) acc1: 100.0000 (76.7847) acc5: 100.0000 (93.1578) time: 0.0128 data: 0.0002 max mem: 29
Test: [27400/50000] eta: 0:05:00 loss: 0.0135 (0.9583) acc1: 100.0000 (76.8038) acc5: 100.0000 (93.1572) time: 0.0129 data: 0.0002 max mem: 29
Test: [27500/50000] eta: 0:04:58 loss: 1.2116 (0.9596) acc1: 0.0000 (76.7863) acc5: 100.0000 (93.1457) time: 0.0125 data: 0.0002 max mem: 29
Test: [27600/50000] eta: 0:04:57 loss: 0.1187 (0.9596) acc1: 100.0000 (76.7834) acc5: 100.0000 (93.1452) time: 0.0123 data: 0.0001 max mem: 29
Test: [27700/50000] eta: 0:04:55 loss: 0.0066 (0.9588) acc1: 100.0000 (76.7987) acc5: 100.0000 (93.1447) time: 0.0127 data: 0.0002 max mem: 29
Test: [27800/50000] eta: 0:04:54 loss: 0.0399 (0.9574) acc1: 100.0000 (76.8426) acc5: 100.0000 (93.1549) time: 0.0128 data: 0.0002 max mem: 29
Test: [27900/50000] eta: 0:04:53 loss: 0.1268 (0.9612) acc1: 100.0000 (76.8073) acc5: 100.0000 (93.1149) time: 0.0129 data: 0.0002 max mem: 29
Test: [28000/50000] eta: 0:04:51 loss: 0.0496 (0.9647) acc1: 100.0000 (76.7223) acc5: 100.0000 (93.0824) time: 0.0127 data: 0.0002 max mem: 29
Test: [28100/50000] eta: 0:04:50 loss: 0.0062 (0.9628) acc1: 100.0000 (76.7695) acc5: 100.0000 (93.0928) time: 0.0128 data: 0.0002 max mem: 29
Test: [28200/50000] eta: 0:04:49 loss: 0.3135 (0.9621) acc1: 100.0000 (76.7810) acc5: 100.0000 (93.0854) time: 0.0146 data: 0.0002 max mem: 29
Test: [28300/50000] eta: 0:04:48 loss: 0.0000 (0.9602) acc1: 100.0000 (76.8135) acc5: 100.0000 (93.1063) time: 0.0145 data: 0.0002 max mem: 29
Test: [28400/50000] eta: 0:04:46 loss: 2.3275 (0.9628) acc1: 0.0000 (76.7508) acc5: 100.0000 (93.0777) time: 0.0145 data: 0.0002 max mem: 29
Test: [28500/50000] eta: 0:04:45 loss: 0.0125 (0.9610) acc1: 100.0000 (76.7903) acc5: 100.0000 (93.0880) time: 0.0127 data: 0.0002 max mem: 29
Test: [28600/50000] eta: 0:04:44 loss: 0.0271 (0.9603) acc1: 100.0000 (76.7945) acc5: 100.0000 (93.0981) time: 0.0126 data: 0.0002 max mem: 29
Test: [28700/50000] eta: 0:04:42 loss: 0.0012 (0.9598) acc1: 100.0000 (76.8092) acc5: 100.0000 (93.1013) time: 0.0125 data: 0.0002 max mem: 29
Test: [28800/50000] eta: 0:04:41 loss: 0.0002 (0.9593) acc1: 100.0000 (76.8272) acc5: 100.0000 (93.1009) time: 0.0137 data: 0.0002 max mem: 29
Test: [28900/50000] eta: 0:04:40 loss: 0.5748 (0.9582) acc1: 100.0000 (76.8209) acc5: 100.0000 (93.1110) time: 0.0140 data: 0.0002 max mem: 29
Test: [29000/50000] eta: 0:04:38 loss: 0.0352 (0.9589) acc1: 100.0000 (76.7801) acc5: 100.0000 (93.1175) time: 0.0145 data: 0.0002 max mem: 29
Test: [29100/50000] eta: 0:04:37 loss: 0.0741 (0.9591) acc1: 100.0000 (76.7912) acc5: 100.0000 (93.1136) time: 0.0128 data: 0.0002 max mem: 29
Test: [29200/50000] eta: 0:04:36 loss: 0.0006 (0.9596) acc1: 100.0000 (76.7919) acc5: 100.0000 (93.1167) time: 0.0144 data: 0.0002 max mem: 29
Test: [29300/50000] eta: 0:04:34 loss: 2.7847 (0.9667) acc1: 0.0000 (76.6766) acc5: 100.0000 (93.0310) time: 0.0142 data: 0.0002 max mem: 29
Test: [29400/50000] eta: 0:04:33 loss: 0.4571 (0.9681) acc1: 100.0000 (76.6403) acc5: 100.0000 (93.0104) time: 0.0124 data: 0.0002 max mem: 29
Test: [29500/50000] eta: 0:04:32 loss: 0.0102 (0.9711) acc1: 100.0000 (76.5838) acc5: 100.0000 (92.9833) time: 0.0126 data: 0.0002 max mem: 29
Test: [29600/50000] eta: 0:04:30 loss: 0.0034 (0.9734) acc1: 100.0000 (76.5413) acc5: 100.0000 (92.9665) time: 0.0126 data: 0.0002 max mem: 29
Test: [29700/50000] eta: 0:04:29 loss: 0.1565 (0.9735) acc1: 100.0000 (76.5462) acc5: 100.0000 (92.9632) time: 0.0125 data: 0.0002 max mem: 29
Test: [29800/50000] eta: 0:04:28 loss: 0.0976 (0.9718) acc1: 100.0000 (76.5813) acc5: 100.0000 (92.9801) time: 0.0125 data: 0.0002 max mem: 29
Test: [29900/50000] eta: 0:04:26 loss: 0.0910 (0.9741) acc1: 100.0000 (76.5392) acc5: 100.0000 (92.9434) time: 0.0124 data: 0.0002 max mem: 29
Test: [30000/50000] eta: 0:04:25 loss: 0.0262 (0.9762) acc1: 100.0000 (76.5008) acc5: 100.0000 (92.9302) time: 0.0123 data: 0.0002 max mem: 29
Test: [30100/50000] eta: 0:04:23 loss: 0.4759 (0.9820) acc1: 100.0000 (76.4061) acc5: 100.0000 (92.8640) time: 0.0126 data: 0.0002 max mem: 29
Test: [30200/50000] eta: 0:04:22 loss: 0.0131 (0.9812) acc1: 100.0000 (76.4346) acc5: 100.0000 (92.8810) time: 0.0124 data: 0.0002 max mem: 29
Test: [30300/50000] eta: 0:04:21 loss: 0.0078 (0.9805) acc1: 100.0000 (76.4595) acc5: 100.0000 (92.8814) time: 0.0123 data: 0.0001 max mem: 29
Test: [30400/50000] eta: 0:04:19 loss: 0.0000 (0.9786) acc1: 100.0000 (76.5074) acc5: 100.0000 (92.8884) time: 0.0123 data: 0.0002 max mem: 29
Test: [30500/50000] eta: 0:04:18 loss: 0.0187 (0.9795) acc1: 100.0000 (76.4991) acc5: 100.0000 (92.8756) time: 0.0122 data: 0.0001 max mem: 29
Test: [30600/50000] eta: 0:04:17 loss: 0.0008 (0.9788) acc1: 100.0000 (76.5433) acc5: 100.0000 (92.8760) time: 0.0124 data: 0.0002 max mem: 29
Test: [30700/50000] eta: 0:04:15 loss: 0.0136 (0.9775) acc1: 100.0000 (76.5675) acc5: 100.0000 (92.8862) time: 0.0129 data: 0.0002 max mem: 29
Test: [30800/50000] eta: 0:04:14 loss: 0.0479 (0.9777) acc1: 100.0000 (76.5722) acc5: 100.0000 (92.8704) time: 0.0130 data: 0.0002 max mem: 29
Test: [30900/50000] eta: 0:04:12 loss: 0.2425 (0.9775) acc1: 100.0000 (76.5865) acc5: 100.0000 (92.8708) time: 0.0124 data: 0.0002 max mem: 29
Test: [31000/50000] eta: 0:04:11 loss: 0.6480 (0.9819) acc1: 100.0000 (76.4879) acc5: 100.0000 (92.8260) time: 0.0123 data: 0.0002 max mem: 29
Test: [31100/50000] eta: 0:04:10 loss: 0.0027 (0.9835) acc1: 100.0000 (76.4059) acc5: 100.0000 (92.8009) time: 0.0123 data: 0.0002 max mem: 29
Test: [31200/50000] eta: 0:04:08 loss: 2.3043 (0.9897) acc1: 0.0000 (76.3020) acc5: 100.0000 (92.7310) time: 0.0122 data: 0.0002 max mem: 29
Test: [31300/50000] eta: 0:04:07 loss: 0.0125 (0.9893) acc1: 100.0000 (76.3011) acc5: 100.0000 (92.7351) time: 0.0124 data: 0.0002 max mem: 29
Test: [31400/50000] eta: 0:04:06 loss: 0.0191 (0.9905) acc1: 100.0000 (76.2778) acc5: 100.0000 (92.7168) time: 0.0126 data: 0.0001 max mem: 29
Test: [31500/50000] eta: 0:04:04 loss: 0.1016 (0.9900) acc1: 100.0000 (76.2896) acc5: 100.0000 (92.7272) time: 0.0129 data: 0.0002 max mem: 29
Test: [31600/50000] eta: 0:04:03 loss: 0.9199 (0.9900) acc1: 100.0000 (76.2856) acc5: 100.0000 (92.7217) time: 0.0125 data: 0.0002 max mem: 29
Test: [31700/50000] eta: 0:04:01 loss: 5.9094 (0.9981) acc1: 0.0000 (76.1774) acc5: 0.0000 (92.6248) time: 0.0125 data: 0.0002 max mem: 29
Test: [31800/50000] eta: 0:04:00 loss: 0.5320 (0.9992) acc1: 100.0000 (76.1580) acc5: 100.0000 (92.6197) time: 0.0126 data: 0.0002 max mem: 29
Test: [31900/50000] eta: 0:03:59 loss: 0.0417 (0.9993) acc1: 100.0000 (76.1387) acc5: 100.0000 (92.6272) time: 0.0143 data: 0.0002 max mem: 29
Test: [32000/50000] eta: 0:03:57 loss: 1.1720 (1.0015) acc1: 0.0000 (75.9976) acc5: 100.0000 (92.6315) time: 0.0125 data: 0.0002 max mem: 29
Test: [32100/50000] eta: 0:03:56 loss: 0.0168 (1.0000) acc1: 100.0000 (76.0350) acc5: 100.0000 (92.6420) time: 0.0133 data: 0.0002 max mem: 29
Test: [32200/50000] eta: 0:03:55 loss: 1.3763 (1.0011) acc1: 0.0000 (75.9976) acc5: 100.0000 (92.6182) time: 0.0126 data: 0.0002 max mem: 29
Test: [32300/50000] eta: 0:03:53 loss: 0.0001 (1.0009) acc1: 100.0000 (76.0193) acc5: 100.0000 (92.6256) time: 0.0131 data: 0.0002 max mem: 29
Test: [32400/50000] eta: 0:03:52 loss: 0.4792 (1.0012) acc1: 100.0000 (76.0100) acc5: 100.0000 (92.6052) time: 0.0125 data: 0.0002 max mem: 29
Test: [32500/50000] eta: 0:03:51 loss: 0.1603 (1.0021) acc1: 100.0000 (76.0100) acc5: 100.0000 (92.5879) time: 0.0134 data: 0.0002 max mem: 29
Test: [32600/50000] eta: 0:03:49 loss: 0.8276 (1.0068) acc1: 0.0000 (75.9271) acc5: 100.0000 (92.5248) time: 0.0145 data: 0.0002 max mem: 29
Test: [32700/50000] eta: 0:03:48 loss: 0.0924 (1.0085) acc1: 100.0000 (75.9029) acc5: 100.0000 (92.4926) time: 0.0144 data: 0.0002 max mem: 29
Test: [32800/50000] eta: 0:03:47 loss: 1.1209 (1.0095) acc1: 100.0000 (75.8940) acc5: 100.0000 (92.4758) time: 0.0122 data: 0.0002 max mem: 29
Test: [32900/50000] eta: 0:03:46 loss: 0.7599 (1.0134) acc1: 0.0000 (75.7940) acc5: 100.0000 (92.4470) time: 0.0124 data: 0.0002 max mem: 29
Test: [33000/50000] eta: 0:03:44 loss: 0.0403 (1.0133) acc1: 100.0000 (75.8007) acc5: 100.0000 (92.4396) time: 0.0121 data: 0.0002 max mem: 29
Test: [33100/50000] eta: 0:03:43 loss: 0.0076 (1.0135) acc1: 100.0000 (75.8104) acc5: 100.0000 (92.4353) time: 0.0121 data: 0.0002 max mem: 29
Test: [33200/50000] eta: 0:03:41 loss: 0.9662 (1.0160) acc1: 0.0000 (75.7537) acc5: 100.0000 (92.4099) time: 0.0122 data: 0.0002 max mem: 29
Test: [33300/50000] eta: 0:03:40 loss: 0.3771 (1.0177) acc1: 100.0000 (75.6884) acc5: 100.0000 (92.4056) time: 0.0122 data: 0.0002 max mem: 29
Test: [33400/50000] eta: 0:03:39 loss: 1.1264 (1.0185) acc1: 0.0000 (75.6654) acc5: 100.0000 (92.3954) time: 0.0122 data: 0.0002 max mem: 29
Test: [33500/50000] eta: 0:03:37 loss: 0.0021 (1.0160) acc1: 100.0000 (75.7231) acc5: 100.0000 (92.4181) time: 0.0121 data: 0.0001 max mem: 29
Test: [33600/50000] eta: 0:03:36 loss: 0.0493 (1.0149) acc1: 100.0000 (75.7269) acc5: 100.0000 (92.4347) time: 0.0121 data: 0.0002 max mem: 29
Test: [33700/50000] eta: 0:03:35 loss: 1.4782 (1.0162) acc1: 0.0000 (75.7010) acc5: 100.0000 (92.4275) time: 0.0122 data: 0.0002 max mem: 29
Test: [33800/50000] eta: 0:03:33 loss: 2.0012 (1.0179) acc1: 0.0000 (75.6664) acc5: 100.0000 (92.4056) time: 0.0122 data: 0.0001 max mem: 29
Test: [33900/50000] eta: 0:03:32 loss: 0.2956 (1.0205) acc1: 100.0000 (75.6261) acc5: 100.0000 (92.3719) time: 0.0121 data: 0.0001 max mem: 29
Test: [34000/50000] eta: 0:03:30 loss: 0.0472 (1.0215) acc1: 100.0000 (75.6272) acc5: 100.0000 (92.3502) time: 0.0123 data: 0.0002 max mem: 29
Test: [34100/50000] eta: 0:03:29 loss: 1.8225 (1.0236) acc1: 0.0000 (75.5755) acc5: 100.0000 (92.3316) time: 0.0122 data: 0.0002 max mem: 29
Test: [34200/50000] eta: 0:03:28 loss: 0.0572 (1.0243) acc1: 100.0000 (75.5709) acc5: 100.0000 (92.3277) time: 0.0122 data: 0.0001 max mem: 29
Test: [34300/50000] eta: 0:03:26 loss: 0.0000 (1.0229) acc1: 100.0000 (75.6100) acc5: 100.0000 (92.3268) time: 0.0124 data: 0.0002 max mem: 29
Test: [34400/50000] eta: 0:03:25 loss: 0.0172 (1.0238) acc1: 100.0000 (75.5908) acc5: 100.0000 (92.3142) time: 0.0122 data: 0.0002 max mem: 29
Test: [34500/50000] eta: 0:03:24 loss: 2.5449 (1.0254) acc1: 0.0000 (75.5572) acc5: 100.0000 (92.2988) time: 0.0121 data: 0.0002 max mem: 29
Test: [34600/50000] eta: 0:03:22 loss: 0.1552 (1.0262) acc1: 100.0000 (75.5556) acc5: 100.0000 (92.2835) time: 0.0121 data: 0.0001 max mem: 29
Test: [34700/50000] eta: 0:03:21 loss: 0.4390 (1.0281) acc1: 100.0000 (75.4935) acc5: 100.0000 (92.2596) time: 0.0121 data: 0.0002 max mem: 29
Test: [34800/50000] eta: 0:03:20 loss: 0.0076 (1.0276) acc1: 100.0000 (75.5151) acc5: 100.0000 (92.2588) time: 0.0122 data: 0.0002 max mem: 29
Test: [34900/50000] eta: 0:03:18 loss: 0.7986 (1.0291) acc1: 100.0000 (75.4792) acc5: 100.0000 (92.2495) time: 0.0121 data: 0.0001 max mem: 29
Test: [35000/50000] eta: 0:03:17 loss: 0.0039 (1.0286) acc1: 100.0000 (75.4921) acc5: 100.0000 (92.2545) time: 0.0123 data: 0.0002 max mem: 29
Test: [35100/50000] eta: 0:03:16 loss: 0.0001 (1.0286) acc1: 100.0000 (75.5021) acc5: 100.0000 (92.2538) time: 0.0122 data: 0.0001 max mem: 29
Test: [35200/50000] eta: 0:03:14 loss: 0.0196 (1.0284) acc1: 100.0000 (75.4922) acc5: 100.0000 (92.2587) time: 0.0123 data: 0.0001 max mem: 29
Test: [35300/50000] eta: 0:03:13 loss: 0.8688 (1.0297) acc1: 0.0000 (75.4766) acc5: 100.0000 (92.2608) time: 0.0122 data: 0.0002 max mem: 29
Test: [35400/50000] eta: 0:03:11 loss: 0.0061 (1.0310) acc1: 100.0000 (75.4668) acc5: 100.0000 (92.2432) time: 0.0122 data: 0.0001 max mem: 29
Test: [35500/50000] eta: 0:03:10 loss: 0.3687 (1.0314) acc1: 100.0000 (75.4373) acc5: 100.0000 (92.2509) time: 0.0121 data: 0.0002 max mem: 29
Test: [35600/50000] eta: 0:03:09 loss: 0.0745 (1.0329) acc1: 100.0000 (75.3996) acc5: 100.0000 (92.2334) time: 0.0124 data: 0.0002 max mem: 29
Test: [35700/50000] eta: 0:03:07 loss: 0.0009 (1.0327) acc1: 100.0000 (75.4153) acc5: 100.0000 (92.2271) time: 0.0125 data: 0.0002 max mem: 29
Test: [35800/50000] eta: 0:03:06 loss: 0.0018 (1.0322) acc1: 100.0000 (75.4309) acc5: 100.0000 (92.2293) time: 0.0121 data: 0.0001 max mem: 29
Test: [35900/50000] eta: 0:03:05 loss: 0.3446 (1.0322) acc1: 100.0000 (75.4352) acc5: 100.0000 (92.2258) time: 0.0122 data: 0.0002 max mem: 29
Test: [36000/50000] eta: 0:03:03 loss: 0.0002 (1.0324) acc1: 100.0000 (75.4312) acc5: 100.0000 (92.2252) time: 0.0122 data: 0.0002 max mem: 29
Test: [36100/50000] eta: 0:03:02 loss: 0.0874 (1.0323) acc1: 100.0000 (75.4411) acc5: 100.0000 (92.2246) time: 0.0121 data: 0.0002 max mem: 29
Test: [36200/50000] eta: 0:03:01 loss: 0.0008 (1.0320) acc1: 100.0000 (75.4620) acc5: 100.0000 (92.2157) time: 0.0121 data: 0.0002 max mem: 29
Test: [36300/50000] eta: 0:02:59 loss: 1.0876 (1.0326) acc1: 0.0000 (75.4332) acc5: 100.0000 (92.2068) time: 0.0119 data: 0.0002 max mem: 29
Test: [36400/50000] eta: 0:02:58 loss: 0.0012 (1.0310) acc1: 100.0000 (75.4677) acc5: 100.0000 (92.2172) time: 0.0123 data: 0.0002 max mem: 29
Test: [36500/50000] eta: 0:02:57 loss: 0.1963 (1.0367) acc1: 100.0000 (75.3925) acc5: 100.0000 (92.1372) time: 0.0123 data: 0.0002 max mem: 29
Test: [36600/50000] eta: 0:02:55 loss: 0.8977 (1.0400) acc1: 0.0000 (75.3367) acc5: 100.0000 (92.1013) time: 0.0122 data: 0.0001 max mem: 29
Test: [36700/50000] eta: 0:02:54 loss: 1.2821 (1.0421) acc1: 100.0000 (75.3031) acc5: 100.0000 (92.0765) time: 0.0129 data: 0.0002 max mem: 29
Test: [36800/50000] eta: 0:02:53 loss: 0.3187 (1.0422) acc1: 100.0000 (75.2860) acc5: 100.0000 (92.0790) time: 0.0130 data: 0.0002 max mem: 29
Test: [36900/50000] eta: 0:02:51 loss: 0.5414 (1.0421) acc1: 100.0000 (75.2852) acc5: 100.0000 (92.0761) time: 0.0126 data: 0.0002 max mem: 29
Test: [37000/50000] eta: 0:02:50 loss: 0.0000 (1.0410) acc1: 100.0000 (75.3142) acc5: 100.0000 (92.0759) time: 0.0124 data: 0.0002 max mem: 29
Test: [37100/50000] eta: 0:02:49 loss: 0.4119 (1.0430) acc1: 100.0000 (75.2594) acc5: 100.0000 (92.0514) time: 0.0123 data: 0.0001 max mem: 29
Test: [37200/50000] eta: 0:02:47 loss: 0.5268 (1.0443) acc1: 100.0000 (75.2292) acc5: 100.0000 (92.0352) time: 0.0124 data: 0.0002 max mem: 29
Test: [37300/50000] eta: 0:02:46 loss: 0.0056 (1.0465) acc1: 100.0000 (75.1588) acc5: 100.0000 (92.0243) time: 0.0123 data: 0.0002 max mem: 29
Test: [37400/50000] eta: 0:02:45 loss: 1.4897 (1.0469) acc1: 0.0000 (75.1611) acc5: 100.0000 (92.0029) time: 0.0122 data: 0.0002 max mem: 29
Test: [37500/50000] eta: 0:02:43 loss: 0.0313 (1.0487) acc1: 100.0000 (75.1100) acc5: 100.0000 (91.9815) time: 0.0123 data: 0.0002 max mem: 29
Test: [37600/50000] eta: 0:02:42 loss: 0.3074 (1.0492) acc1: 100.0000 (75.0911) acc5: 100.0000 (91.9869) time: 0.0119 data: 0.0001 max mem: 29
Test: [37700/50000] eta: 0:02:41 loss: 0.0172 (1.0496) acc1: 100.0000 (75.0882) acc5: 100.0000 (91.9790) time: 0.0119 data: 0.0002 max mem: 29
Test: [37800/50000] eta: 0:02:39 loss: 0.0022 (1.0492) acc1: 100.0000 (75.0985) acc5: 100.0000 (91.9817) time: 0.0119 data: 0.0002 max mem: 29
Test: [37900/50000] eta: 0:02:38 loss: 0.0170 (1.0489) acc1: 100.0000 (75.1036) acc5: 100.0000 (91.9765) time: 0.0119 data: 0.0001 max mem: 29
Test: [38000/50000] eta: 0:02:37 loss: 0.0966 (1.0499) acc1: 100.0000 (75.0954) acc5: 100.0000 (91.9502) time: 0.0120 data: 0.0002 max mem: 29
Test: [38100/50000] eta: 0:02:35 loss: 0.0094 (1.0497) acc1: 100.0000 (75.0978) acc5: 100.0000 (91.9556) time: 0.0120 data: 0.0002 max mem: 29
Test: [38200/50000] eta: 0:02:34 loss: 0.0218 (1.0517) acc1: 100.0000 (75.0792) acc5: 100.0000 (91.9191) time: 0.0119 data: 0.0001 max mem: 29
Test: [38300/50000] eta: 0:02:33 loss: 0.0913 (1.0534) acc1: 100.0000 (75.0503) acc5: 100.0000 (91.9010) time: 0.0120 data: 0.0001 max mem: 29
Test: [38400/50000] eta: 0:02:31 loss: 3.0212 (1.0541) acc1: 0.0000 (75.0475) acc5: 100.0000 (91.8830) time: 0.0120 data: 0.0002 max mem: 29
Test: [38500/50000] eta: 0:02:30 loss: 0.0590 (1.0545) acc1: 100.0000 (75.0500) acc5: 100.0000 (91.8677) time: 0.0119 data: 0.0002 max mem: 29
Test: [38600/50000] eta: 0:02:28 loss: 0.0100 (1.0544) acc1: 100.0000 (75.0706) acc5: 100.0000 (91.8681) time: 0.0118 data: 0.0002 max mem: 29
Test: [38700/50000] eta: 0:02:27 loss: 0.4904 (1.0564) acc1: 100.0000 (75.0239) acc5: 100.0000 (91.8478) time: 0.0119 data: 0.0002 max mem: 29
Test: [38800/50000] eta: 0:02:26 loss: 0.4960 (1.0580) acc1: 100.0000 (74.9852) acc5: 100.0000 (91.8250) time: 0.0119 data: 0.0002 max mem: 29
Test: [38900/50000] eta: 0:02:24 loss: 0.0769 (1.0580) acc1: 100.0000 (74.9852) acc5: 100.0000 (91.8305) time: 0.0120 data: 0.0001 max mem: 29
Test: [39000/50000] eta: 0:02:23 loss: 0.0051 (1.0587) acc1: 100.0000 (74.9904) acc5: 100.0000 (91.8207) time: 0.0120 data: 0.0002 max mem: 29
Test: [39100/50000] eta: 0:02:22 loss: 0.0018 (1.0566) acc1: 100.0000 (75.0313) acc5: 100.0000 (91.8391) time: 0.0119 data: 0.0002 max mem: 29
Test: [39200/50000] eta: 0:02:20 loss: 0.0012 (1.0582) acc1: 100.0000 (74.9802) acc5: 100.0000 (91.8242) time: 0.0119 data: 0.0002 max mem: 29
Test: [39300/50000] eta: 0:02:19 loss: 0.1853 (1.0605) acc1: 100.0000 (74.9396) acc5: 100.0000 (91.7865) time: 0.0119 data: 0.0001 max mem: 29
Test: [39400/50000] eta: 0:02:18 loss: 0.6406 (1.0621) acc1: 0.0000 (74.9042) acc5: 100.0000 (91.7667) time: 0.0119 data: 0.0002 max mem: 29
Test: [39500/50000] eta: 0:02:16 loss: 0.1936 (1.0617) acc1: 100.0000 (74.9095) acc5: 100.0000 (91.7698) time: 0.0119 data: 0.0002 max mem: 29
Test: [39600/50000] eta: 0:02:15 loss: 0.0425 (1.0628) acc1: 100.0000 (74.8946) acc5: 100.0000 (91.7527) time: 0.0120 data: 0.0002 max mem: 29
Test: [39700/50000] eta: 0:02:14 loss: 0.0153 (1.0645) acc1: 100.0000 (74.8747) acc5: 100.0000 (91.7357) time: 0.0121 data: 0.0002 max mem: 29
Test: [39800/50000] eta: 0:02:12 loss: 0.0769 (1.0651) acc1: 100.0000 (74.8675) acc5: 100.0000 (91.7238) time: 0.0119 data: 0.0002 max mem: 29
Test: [39900/50000] eta: 0:02:11 loss: 0.0277 (1.0658) acc1: 100.0000 (74.8528) acc5: 100.0000 (91.7170) time: 0.0120 data: 0.0002 max mem: 29
Test: [40000/50000] eta: 0:02:10 loss: 0.6600 (1.0687) acc1: 100.0000 (74.8131) acc5: 100.0000 (91.6777) time: 0.0120 data: 0.0002 max mem: 29
Test: [40100/50000] eta: 0:02:08 loss: 0.0090 (1.0668) acc1: 100.0000 (74.8560) acc5: 100.0000 (91.6935) time: 0.0121 data: 0.0002 max mem: 29
Test: [40200/50000] eta: 0:02:07 loss: 0.0264 (1.0652) acc1: 100.0000 (74.8887) acc5: 100.0000 (91.7067) time: 0.0120 data: 0.0002 max mem: 29
Test: [40300/50000] eta: 0:02:06 loss: 0.0093 (1.0654) acc1: 100.0000 (74.8815) acc5: 100.0000 (91.7099) time: 0.0120 data: 0.0002 max mem: 29
Test: [40400/50000] eta: 0:02:04 loss: 0.0218 (1.0666) acc1: 100.0000 (74.8744) acc5: 100.0000 (91.6834) time: 0.0121 data: 0.0002 max mem: 29
Test: [40500/50000] eta: 0:02:03 loss: 1.6256 (1.0685) acc1: 0.0000 (74.8228) acc5: 100.0000 (91.6644) time: 0.0120 data: 0.0002 max mem: 29
Test: [40600/50000] eta: 0:02:02 loss: 0.1739 (1.0713) acc1: 100.0000 (74.7691) acc5: 100.0000 (91.6381) time: 0.0120 data: 0.0002 max mem: 29
Test: [40700/50000] eta: 0:02:01 loss: 2.0721 (1.0730) acc1: 0.0000 (74.7353) acc5: 100.0000 (91.6145) time: 0.0121 data: 0.0002 max mem: 29
Test: [40800/50000] eta: 0:01:59 loss: 0.2095 (1.0723) acc1: 100.0000 (74.7555) acc5: 100.0000 (91.6277) time: 0.0121 data: 0.0002 max mem: 29
Test: [40900/50000] eta: 0:01:58 loss: 0.9667 (1.0716) acc1: 100.0000 (74.7610) acc5: 100.0000 (91.6359) time: 0.0121 data: 0.0002 max mem: 29
Test: [41000/50000] eta: 0:01:57 loss: 0.1690 (1.0745) acc1: 100.0000 (74.7128) acc5: 100.0000 (91.6051) time: 0.0122 data: 0.0002 max mem: 29
Test: [41100/50000] eta: 0:01:55 loss: 0.0553 (1.0729) acc1: 100.0000 (74.7500) acc5: 100.0000 (91.6182) time: 0.0120 data: 0.0001 max mem: 29
Test: [41200/50000] eta: 0:01:54 loss: 0.7887 (1.0736) acc1: 100.0000 (74.7506) acc5: 100.0000 (91.5973) time: 0.0120 data: 0.0002 max mem: 29
Test: [41300/50000] eta: 0:01:53 loss: 0.0577 (1.0742) acc1: 100.0000 (74.7125) acc5: 100.0000 (91.5958) time: 0.0120 data: 0.0002 max mem: 29
Test: [41400/50000] eta: 0:01:51 loss: 0.4766 (1.0766) acc1: 100.0000 (74.6649) acc5: 100.0000 (91.5751) time: 0.0119 data: 0.0002 max mem: 29
Test: [41500/50000] eta: 0:01:50 loss: 0.0224 (1.0772) acc1: 100.0000 (74.6633) acc5: 100.0000 (91.5616) time: 0.0120 data: 0.0001 max mem: 29
Test: [41600/50000] eta: 0:01:49 loss: 0.2120 (1.0775) acc1: 100.0000 (74.6569) acc5: 100.0000 (91.5579) time: 0.0120 data: 0.0002 max mem: 29
Test: [41700/50000] eta: 0:01:47 loss: 0.0137 (1.0766) acc1: 100.0000 (74.6745) acc5: 100.0000 (91.5661) time: 0.0119 data: 0.0002 max mem: 29
Test: [41800/50000] eta: 0:01:46 loss: 0.0237 (1.0781) acc1: 0.0000 (74.6322) acc5: 100.0000 (91.5409) time: 0.0120 data: 0.0002 max mem: 29
Test: [41900/50000] eta: 0:01:45 loss: 2.3601 (1.0831) acc1: 0.0000 (74.5113) acc5: 100.0000 (91.4775) time: 0.0120 data: 0.0002 max mem: 29
Test: [42000/50000] eta: 0:01:43 loss: 0.3241 (1.0853) acc1: 100.0000 (74.4601) acc5: 100.0000 (91.4597) time: 0.0121 data: 0.0001 max mem: 29
Test: [42100/50000] eta: 0:01:42 loss: 0.9063 (1.0870) acc1: 100.0000 (74.4329) acc5: 100.0000 (91.4349) time: 0.0121 data: 0.0002 max mem: 29
Test: [42200/50000] eta: 0:01:41 loss: 0.0135 (1.0873) acc1: 100.0000 (74.4129) acc5: 100.0000 (91.4291) time: 0.0119 data: 0.0001 max mem: 29
Test: [42300/50000] eta: 0:01:39 loss: 0.1794 (1.0892) acc1: 100.0000 (74.3883) acc5: 100.0000 (91.4045) time: 0.0120 data: 0.0002 max mem: 29
Test: [42400/50000] eta: 0:01:38 loss: 0.0941 (1.0898) acc1: 100.0000 (74.3615) acc5: 100.0000 (91.3988) time: 0.0119 data: 0.0002 max mem: 29
Test: [42500/50000] eta: 0:01:37 loss: 0.0792 (1.0902) acc1: 100.0000 (74.3347) acc5: 100.0000 (91.4073) time: 0.0121 data: 0.0002 max mem: 29
Test: [42600/50000] eta: 0:01:35 loss: 0.0957 (1.0900) acc1: 100.0000 (74.3292) acc5: 100.0000 (91.4133) time: 0.0120 data: 0.0002 max mem: 29
Test: [42700/50000] eta: 0:01:34 loss: 0.0044 (1.0891) acc1: 100.0000 (74.3472) acc5: 100.0000 (91.4217) time: 0.0121 data: 0.0002 max mem: 29
Test: [42800/50000] eta: 0:01:33 loss: 0.0020 (1.0896) acc1: 100.0000 (74.3487) acc5: 100.0000 (91.4091) time: 0.0122 data: 0.0002 max mem: 29
Test: [42900/50000] eta: 0:01:32 loss: 0.2166 (1.0903) acc1: 100.0000 (74.3293) acc5: 100.0000 (91.4105) time: 0.0120 data: 0.0002 max mem: 29
Test: [43000/50000] eta: 0:01:30 loss: 0.4386 (1.0907) acc1: 100.0000 (74.3239) acc5: 100.0000 (91.3909) time: 0.0121 data: 0.0002 max mem: 29
Test: [43100/50000] eta: 0:01:29 loss: 0.1530 (1.0923) acc1: 100.0000 (74.2883) acc5: 100.0000 (91.3668) time: 0.0121 data: 0.0002 max mem: 29
Test: [43200/50000] eta: 0:01:28 loss: 0.0065 (1.0916) acc1: 100.0000 (74.3085) acc5: 100.0000 (91.3706) time: 0.0121 data: 0.0002 max mem: 29
Test: [43300/50000] eta: 0:01:26 loss: 0.5499 (1.0922) acc1: 100.0000 (74.2685) acc5: 100.0000 (91.3651) time: 0.0124 data: 0.0002 max mem: 29
Test: [43400/50000] eta: 0:01:25 loss: 0.3370 (1.0916) acc1: 100.0000 (74.2748) acc5: 100.0000 (91.3735) time: 0.0121 data: 0.0002 max mem: 29
Test: [43500/50000] eta: 0:01:24 loss: 0.2863 (1.0939) acc1: 100.0000 (74.2374) acc5: 100.0000 (91.3519) time: 0.0121 data: 0.0002 max mem: 29
Test: [43600/50000] eta: 0:01:22 loss: 0.2812 (1.0930) acc1: 100.0000 (74.2391) acc5: 100.0000 (91.3672) time: 0.0122 data: 0.0002 max mem: 29
Test: [43700/50000] eta: 0:01:21 loss: 0.0013 (1.0925) acc1: 100.0000 (74.2592) acc5: 100.0000 (91.3663) time: 0.0123 data: 0.0002 max mem: 29
Test: [43800/50000] eta: 0:01:20 loss: 0.1880 (1.0916) acc1: 100.0000 (74.2791) acc5: 100.0000 (91.3723) time: 0.0121 data: 0.0002 max mem: 29
Test: [43900/50000] eta: 0:01:18 loss: 0.0064 (1.0920) acc1: 100.0000 (74.2375) acc5: 100.0000 (91.3692) time: 0.0123 data: 0.0001 max mem: 29
Test: [44000/50000] eta: 0:01:17 loss: 0.0775 (1.0924) acc1: 100.0000 (74.2165) acc5: 100.0000 (91.3638) time: 0.0122 data: 0.0002 max mem: 29
Test: [44100/50000] eta: 0:01:16 loss: 0.1228 (1.0920) acc1: 100.0000 (74.2341) acc5: 100.0000 (91.3675) time: 0.0122 data: 0.0002 max mem: 29
Test: [44200/50000] eta: 0:01:15 loss: 1.8513 (1.0932) acc1: 0.0000 (74.2110) acc5: 100.0000 (91.3509) time: 0.0122 data: 0.0002 max mem: 29
Test: [44300/50000] eta: 0:01:13 loss: 3.8736 (1.0965) acc1: 0.0000 (74.1541) acc5: 100.0000 (91.3117) time: 0.0122 data: 0.0001 max mem: 29
Test: [44400/50000] eta: 0:01:12 loss: 0.4045 (1.0968) acc1: 100.0000 (74.1650) acc5: 100.0000 (91.3065) time: 0.0122 data: 0.0002 max mem: 29
Test: [44500/50000] eta: 0:01:11 loss: 0.4718 (1.0967) acc1: 100.0000 (74.1579) acc5: 100.0000 (91.3036) time: 0.0121 data: 0.0002 max mem: 29
Test: [44600/50000] eta: 0:01:09 loss: 0.0673 (1.0961) acc1: 100.0000 (74.1732) acc5: 100.0000 (91.3051) time: 0.0124 data: 0.0002 max mem: 29
Test: [44700/50000] eta: 0:01:08 loss: 1.4266 (1.0967) acc1: 0.0000 (74.1370) acc5: 100.0000 (91.3089) time: 0.0122 data: 0.0002 max mem: 29
Test: [44800/50000] eta: 0:01:07 loss: 0.0431 (1.0960) acc1: 100.0000 (74.1479) acc5: 100.0000 (91.3216) time: 0.0134 data: 0.0002 max mem: 29
Test: [44900/50000] eta: 0:01:05 loss: 0.0027 (1.0956) acc1: 100.0000 (74.1632) acc5: 100.0000 (91.3276) time: 0.0121 data: 0.0002 max mem: 29
Test: [45000/50000] eta: 0:01:04 loss: 2.0405 (1.0988) acc1: 0.0000 (74.0917) acc5: 100.0000 (91.3024) time: 0.0137 data: 0.0002 max mem: 29
Test: [45100/50000] eta: 0:01:03 loss: 0.0499 (1.0986) acc1: 100.0000 (74.1026) acc5: 100.0000 (91.2907) time: 0.0122 data: 0.0002 max mem: 29
Test: [45200/50000] eta: 0:01:02 loss: 0.0301 (1.0984) acc1: 100.0000 (74.1112) acc5: 100.0000 (91.2944) time: 0.0137 data: 0.0002 max mem: 29
Test: [45300/50000] eta: 0:01:00 loss: 0.2390 (1.0996) acc1: 100.0000 (74.0889) acc5: 100.0000 (91.2629) time: 0.0120 data: 0.0002 max mem: 29
Test: [45400/50000] eta: 0:00:59 loss: 1.2460 (1.1034) acc1: 0.0000 (74.0006) acc5: 100.0000 (91.2227) time: 0.0121 data: 0.0002 max mem: 29
Test: [45500/50000] eta: 0:00:58 loss: 0.6810 (1.1048) acc1: 100.0000 (73.9588) acc5: 100.0000 (91.2266) time: 0.0121 data: 0.0002 max mem: 29
Test: [45600/50000] eta: 0:00:56 loss: 0.2011 (1.1071) acc1: 100.0000 (73.9194) acc5: 100.0000 (91.2041) time: 0.0120 data: 0.0002 max mem: 29
Test: [45700/50000] eta: 0:00:55 loss: 0.0228 (1.1068) acc1: 100.0000 (73.9240) acc5: 100.0000 (91.2059) time: 0.0121 data: 0.0001 max mem: 29
Test: [45800/50000] eta: 0:00:54 loss: 0.0003 (1.1058) acc1: 100.0000 (73.9416) acc5: 100.0000 (91.2185) time: 0.0122 data: 0.0002 max mem: 29
Test: [45900/50000] eta: 0:00:52 loss: 0.1972 (1.1048) acc1: 100.0000 (73.9657) acc5: 100.0000 (91.2289) time: 0.0121 data: 0.0002 max mem: 29
Test: [46000/50000] eta: 0:00:51 loss: 0.1110 (1.1049) acc1: 100.0000 (73.9745) acc5: 100.0000 (91.2176) time: 0.0122 data: 0.0002 max mem: 29
Test: [46100/50000] eta: 0:00:50 loss: 0.6403 (1.1047) acc1: 100.0000 (73.9745) acc5: 100.0000 (91.2236) time: 0.0121 data: 0.0002 max mem: 29
Test: [46200/50000] eta: 0:00:49 loss: 1.0495 (1.1057) acc1: 100.0000 (73.9703) acc5: 100.0000 (91.2145) time: 0.0122 data: 0.0002 max mem: 29
Test: [46300/50000] eta: 0:00:47 loss: 0.4804 (1.1053) acc1: 100.0000 (73.9595) acc5: 100.0000 (91.2183) time: 0.0120 data: 0.0001 max mem: 29
Test: [46400/50000] eta: 0:00:46 loss: 0.0027 (1.1038) acc1: 100.0000 (73.9812) acc5: 100.0000 (91.2351) time: 0.0121 data: 0.0002 max mem: 29
Test: [46500/50000] eta: 0:00:45 loss: 0.0935 (1.1043) acc1: 100.0000 (73.9575) acc5: 100.0000 (91.2346) time: 0.0121 data: 0.0001 max mem: 29
Test: [46600/50000] eta: 0:00:43 loss: 0.0400 (1.1054) acc1: 100.0000 (73.9297) acc5: 100.0000 (91.2255) time: 0.0120 data: 0.0001 max mem: 29
Test: [46700/50000] eta: 0:00:42 loss: 0.0019 (1.1049) acc1: 100.0000 (73.9406) acc5: 100.0000 (91.2207) time: 0.0121 data: 0.0001 max mem: 29
Test: [46800/50000] eta: 0:00:41 loss: 0.0435 (1.1051) acc1: 100.0000 (73.9429) acc5: 100.0000 (91.2160) time: 0.0120 data: 0.0002 max mem: 29
Test: [46900/50000] eta: 0:00:40 loss: 0.0004 (1.1039) acc1: 100.0000 (73.9707) acc5: 100.0000 (91.2283) time: 0.0126 data: 0.0002 max mem: 29
Test: [47000/50000] eta: 0:00:38 loss: 0.0368 (1.1030) acc1: 100.0000 (73.9771) acc5: 100.0000 (91.2470) time: 0.0124 data: 0.0001 max mem: 29
Test: [47100/50000] eta: 0:00:37 loss: 0.0096 (1.1024) acc1: 100.0000 (73.9814) acc5: 100.0000 (91.2550) time: 0.0126 data: 0.0002 max mem: 29
Test: [47200/50000] eta: 0:00:36 loss: 0.1077 (1.1020) acc1: 100.0000 (73.9772) acc5: 100.0000 (91.2608) time: 0.0139 data: 0.0002 max mem: 29
Test: [47300/50000] eta: 0:00:34 loss: 0.0041 (1.1010) acc1: 100.0000 (73.9984) acc5: 100.0000 (91.2729) time: 0.0134 data: 0.0002 max mem: 29
Test: [47400/50000] eta: 0:00:33 loss: 0.6002 (1.1012) acc1: 100.0000 (73.9816) acc5: 100.0000 (91.2871) time: 0.0127 data: 0.0002 max mem: 29
Test: [47500/50000] eta: 0:00:32 loss: 0.0593 (1.1010) acc1: 100.0000 (73.9816) acc5: 100.0000 (91.2907) time: 0.0123 data: 0.0002 max mem: 29
Test: [47600/50000] eta: 0:00:30 loss: 0.1147 (1.1010) acc1: 100.0000 (73.9795) acc5: 100.0000 (91.2985) time: 0.0124 data: 0.0002 max mem: 29
Test: [47700/50000] eta: 0:00:29 loss: 0.0013 (1.0998) acc1: 100.0000 (74.0068) acc5: 100.0000 (91.3042) time: 0.0126 data: 0.0002 max mem: 29
Test: [47800/50000] eta: 0:00:28 loss: 0.0000 (1.0988) acc1: 100.0000 (74.0298) acc5: 100.0000 (91.3140) time: 0.0124 data: 0.0002 max mem: 29
Test: [47900/50000] eta: 0:00:27 loss: 0.0004 (1.0977) acc1: 100.0000 (74.0590) acc5: 100.0000 (91.3259) time: 0.0121 data: 0.0002 max mem: 29
Test: [48000/50000] eta: 0:00:25 loss: 0.0048 (1.0966) acc1: 100.0000 (74.0818) acc5: 100.0000 (91.3335) time: 0.0121 data: 0.0002 max mem: 29
Test: [48100/50000] eta: 0:00:24 loss: 1.0288 (1.0997) acc1: 100.0000 (74.0276) acc5: 100.0000 (91.2996) time: 0.0123 data: 0.0002 max mem: 29
Test: [48200/50000] eta: 0:00:23 loss: 0.0181 (1.0999) acc1: 100.0000 (74.0296) acc5: 100.0000 (91.2927) time: 0.0122 data: 0.0002 max mem: 29
Test: [48300/50000] eta: 0:00:21 loss: 0.0095 (1.0998) acc1: 100.0000 (74.0399) acc5: 100.0000 (91.2921) time: 0.0125 data: 0.0002 max mem: 29
Test: [48400/50000] eta: 0:00:20 loss: 0.1439 (1.1010) acc1: 100.0000 (74.0171) acc5: 100.0000 (91.2874) time: 0.0121 data: 0.0002 max mem: 29
Test: [48500/50000] eta: 0:00:19 loss: 2.2339 (1.1049) acc1: 0.0000 (73.9449) acc5: 100.0000 (91.2435) time: 0.0121 data: 0.0002 max mem: 29
Test: [48600/50000] eta: 0:00:18 loss: 0.0188 (1.1055) acc1: 100.0000 (73.9244) acc5: 100.0000 (91.2409) time: 0.0121 data: 0.0001 max mem: 29
Test: [48700/50000] eta: 0:00:16 loss: 0.1100 (1.1054) acc1: 100.0000 (73.9163) acc5: 100.0000 (91.2527) time: 0.0122 data: 0.0002 max mem: 29
Test: [48800/50000] eta: 0:00:15 loss: 1.7498 (1.1053) acc1: 0.0000 (73.9206) acc5: 100.0000 (91.2584) time: 0.0121 data: 0.0002 max mem: 29
Test: [48900/50000] eta: 0:00:14 loss: 0.3690 (1.1059) acc1: 100.0000 (73.8983) acc5: 100.0000 (91.2660) time: 0.0120 data: 0.0002 max mem: 29
Test: [49000/50000] eta: 0:00:12 loss: 0.1133 (1.1072) acc1: 100.0000 (73.8699) acc5: 100.0000 (91.2573) time: 0.0120 data: 0.0002 max mem: 29
Test: [49100/50000] eta: 0:00:11 loss: 0.1942 (1.1067) acc1: 100.0000 (73.8906) acc5: 100.0000 (91.2609) time: 0.0122 data: 0.0002 max mem: 29
Test: [49200/50000] eta: 0:00:10 loss: 0.0112 (1.1064) acc1: 100.0000 (73.9030) acc5: 100.0000 (91.2624) time: 0.0132 data: 0.0002 max mem: 29
Test: [49300/50000] eta: 0:00:09 loss: 0.0015 (1.1046) acc1: 100.0000 (73.9478) acc5: 100.0000 (91.2760) time: 0.0121 data: 0.0002 max mem: 29
Test: [49400/50000] eta: 0:00:07 loss: 0.6341 (1.1039) acc1: 0.0000 (73.9459) acc5: 100.0000 (91.2897) time: 0.0120 data: 0.0001 max mem: 29
Test: [49500/50000] eta: 0:00:06 loss: 0.0015 (1.1020) acc1: 100.0000 (73.9884) acc5: 100.0000 (91.3052) time: 0.0120 data: 0.0002 max mem: 29
Test: [49600/50000] eta: 0:00:05 loss: 0.0005 (1.1002) acc1: 100.0000 (74.0287) acc5: 100.0000 (91.3207) time: 0.0120 data: 0.0001 max mem: 29
Test: [49700/50000] eta: 0:00:03 loss: 0.0000 (1.0989) acc1: 100.0000 (74.0629) acc5: 100.0000 (91.3342) time: 0.0120 data: 0.0001 max mem: 29
Test: [49800/50000] eta: 0:00:02 loss: 0.0000 (1.0973) acc1: 100.0000 (74.1049) acc5: 100.0000 (91.3496) time: 0.0120 data: 0.0001 max mem: 29
Test: [49900/50000] eta: 0:00:01 loss: 0.5711 (1.0966) acc1: 100.0000 (74.1188) acc5: 100.0000 (91.3589) time: 0.0120 data: 0.0002 max mem: 29
Test: Total time: 0:10:44
* Acc@1 74.054 Acc@5 91.340
tvm.result.log 0 → 100644
View file @ cea8529d
[root@dcunode5 mobilenet]# python val_onnx.py --test-only --data-path /parastor/DL_DATA/ImageNet-pytorch --model mobilenet_v3_large --b 1 --pretrained
Not using distributed mode
Namespace(apex=False, apex_opt_level='O1', auto_augment=None, batch_size=1, cache_dataset=False, data_path='/parastor/DL_DATA/ImageNet-pytorch', device='cuda', dist_url='env://', distributed=False, epochs=90, lr=0.1, lr_gamma=0.1, lr_step_size=30, model='mobilenet_v3_large', momentum=0.9, opt='sgd', output_dir='.', pretrained=True, print_freq=10, random_erase=0.0, resume='', start_epoch=0, sync_bn=False, test_only=True, weight_decay=0.0001, workers=16, world_size=1)
Loading data
Loading training data
Took 314.8946807384491
Loading validation data
Creating data loaders
Creating model
/usr/local/lib/python3.8/site-packages/torch/onnx/utils.py:88: UserWarning: `enable_onnx_checker' is deprecated and ignored. It will be removed inthe next PyTorch release. To proceed despite ONNX checker failures, youcan catch torch.onnx.ONNXCheckerError.
warnings.warn("`enable_onnx_checker' is deprecated and ignored. It will be removed in"
Compile...
[02:31:32] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03568 ms, Memory: 1354752
One or more operators have not been tuned. Please tune your model for better performance. Use DEBUG logging level to see more details.
[02:31:32] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02656 ms, Memory: 0
[02:31:32] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02768 ms, Memory: 0
[02:31:32] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03136 ms, Memory: 0
[02:31:32] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02448 ms, Memory: 0
[02:31:32] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03488 ms, Memory: 0
[02:31:32] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02448 ms, Memory: 0
[02:31:32] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.01584 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.01264 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.04288 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.01776 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.01456 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03024 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.01776 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.01456 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03024 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.0232 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02736 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02496 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.01888 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.0496 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.01888 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.0496 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.0176 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.05488 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02256 ms, Memory: 0
[02:31:33] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.04 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.03456 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03632 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.04 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.03456 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.031198 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.05568 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.040478 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.031198 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.05568 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.040478 ms, Memory: 0
[02:31:34] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.031198 ms, Memory: 0
[02:31:53] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03568 ms, Memory: 1354752
[02:31:53] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02656 ms, Memory: 0
[02:31:53] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02768 ms, Memory: 0
[02:31:54] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03136 ms, Memory: 0
[02:31:54] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02448 ms, Memory: 0
[02:31:54] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03488 ms, Memory: 0
[02:31:55] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.01584 ms, Memory: 0
[02:31:55] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.01264 ms, Memory: 0
[02:31:55] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.04288 ms, Memory: 0
[02:31:55] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.01776 ms, Memory: 0
[02:31:56] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:56] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.01456 ms, Memory: 0
[02:31:56] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03024 ms, Memory: 0
[02:31:56] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.0232 ms, Memory: 0
[02:31:57] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02736 ms, Memory: 0
[02:31:57] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:57] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02496 ms, Memory: 0
[02:31:57] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.01888 ms, Memory: 0
[02:31:58] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.0496 ms, Memory: 0
[02:31:58] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.0176 ms, Memory: 0
[02:31:59] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.05488 ms, Memory: 0
[02:31:59] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.02256 ms, Memory: 0
[02:31:59] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:31:59] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.04 ms, Memory: 0
[02:32:00] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:32:00] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.03456 ms, Memory: 0
[02:32:00] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.03632 ms, Memory: 0
[02:32:01] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:32:01] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.031198 ms, Memory: 0
[02:32:02] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0 ms, Memory: 0
[02:32:02] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoDirect - time: 0.05568 ms, Memory: 0
[02:32:02] /root/tvm-0.11-dev0/src/runtime/contrib/miopen/conv_forward.cc:156: 0) miopenConvolutionFwdAlgoGEMM - time: 0.040478 ms, Memory: 0
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
[02:32:03] /root/tvm-0.11-dev0/src/target/llvm/codegen_llvm.cc:1796: Warning: Unroll hint get ignore at CodeGenLLVM backend, consider set unroll_explicit=True
Compile success!
Test: [ 0/50000] eta: 21:13:14 loss: 3.7605 (3.7605) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 1.5279 data: 1.2693 max mem: 86
Test: [ 100/50000] eta: 0:16:33 loss: 0.0005 (0.5272) acc1: 100.0000 (88.1188) acc5: 100.0000 (97.0297) time: 0.0051 data: 0.0002 max mem: 86
Test: [ 200/50000] eta: 0:10:23 loss: 0.1722 (0.6195) acc1: 100.0000 (82.5871) acc5: 100.0000 (96.5174) time: 0.0050 data: 0.0002 max mem: 86
Test: [ 300/50000] eta: 0:08:11 loss: 0.0325 (0.6624) acc1: 100.0000 (82.0598) acc5: 100.0000 (96.0133) time: 0.0046 data: 0.0002 max mem: 86
Test: [ 400/50000] eta: 0:07:03 loss: 0.0283 (0.7113) acc1: 100.0000 (81.7955) acc5: 100.0000 (94.7631) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 500/50000] eta: 0:06:19 loss: 0.0000 (0.6293) acc1: 100.0000 (83.8323) acc5: 100.0000 (95.4092) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 600/50000] eta: 0:05:51 loss: 0.0002 (0.5730) acc1: 100.0000 (85.3577) acc5: 100.0000 (95.8403) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 700/50000] eta: 0:05:30 loss: 0.0023 (0.5215) acc1: 100.0000 (86.5906) acc5: 100.0000 (96.2910) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 800/50000] eta: 0:05:13 loss: 0.0003 (0.4884) acc1: 100.0000 (87.5156) acc5: 100.0000 (96.5044) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 900/50000] eta: 0:04:59 loss: 0.0004 (0.4812) acc1: 100.0000 (87.6804) acc5: 100.0000 (96.8923) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 1000/50000] eta: 0:04:50 loss: 0.0004 (0.4654) acc1: 100.0000 (88.4116) acc5: 100.0000 (97.0030) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 1100/50000] eta: 0:04:42 loss: 0.0158 (0.5022) acc1: 100.0000 (87.4659) acc5: 100.0000 (96.5486) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 1200/50000] eta: 0:04:35 loss: 0.0051 (0.4699) acc1: 100.0000 (88.1765) acc5: 100.0000 (96.8360) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 1300/50000] eta: 0:04:30 loss: 0.0049 (0.4499) acc1: 100.0000 (88.8547) acc5: 100.0000 (96.9254) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 1400/50000] eta: 0:04:24 loss: 0.0190 (0.5285) acc1: 100.0000 (87.2234) acc5: 100.0000 (96.4311) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 1500/50000] eta: 0:04:19 loss: 0.0029 (0.5403) acc1: 100.0000 (87.1419) acc5: 100.0000 (96.1359) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 1600/50000] eta: 0:04:16 loss: 0.0291 (0.5636) acc1: 100.0000 (86.8832) acc5: 100.0000 (96.0025) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 1700/50000] eta: 0:04:13 loss: 0.0790 (0.6317) acc1: 100.0000 (85.3028) acc5: 100.0000 (95.7084) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 1800/50000] eta: 0:04:10 loss: 0.8301 (0.6861) acc1: 0.0000 (83.8978) acc5: 100.0000 (95.2804) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 1900/50000] eta: 0:04:07 loss: 0.1593 (0.6977) acc1: 100.0000 (83.2194) acc5: 100.0000 (95.4235) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 2000/50000] eta: 0:04:05 loss: 0.0165 (0.7040) acc1: 100.0000 (83.2084) acc5: 100.0000 (95.4023) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 2100/50000] eta: 0:04:03 loss: 0.0156 (0.7113) acc1: 100.0000 (82.8653) acc5: 100.0000 (95.4307) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 2200/50000] eta: 0:04:01 loss: 0.0035 (0.7085) acc1: 100.0000 (82.9623) acc5: 100.0000 (95.4112) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 2300/50000] eta: 0:03:59 loss: 0.0000 (0.7477) acc1: 100.0000 (82.3990) acc5: 100.0000 (95.0456) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 2400/50000] eta: 0:03:57 loss: 0.0144 (0.7674) acc1: 100.0000 (81.7576) acc5: 100.0000 (94.9188) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 2500/50000] eta: 0:03:55 loss: 0.1349 (0.7727) acc1: 100.0000 (81.7673) acc5: 100.0000 (94.7621) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 2600/50000] eta: 0:03:53 loss: 0.0001 (0.7679) acc1: 100.0000 (81.8916) acc5: 100.0000 (94.6943) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 2700/50000] eta: 0:03:50 loss: 0.2303 (0.7688) acc1: 100.0000 (81.7845) acc5: 100.0000 (94.6686) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 2800/50000] eta: 0:03:48 loss: 0.1836 (0.7950) acc1: 100.0000 (81.1853) acc5: 100.0000 (94.5734) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 2900/50000] eta: 0:03:47 loss: 0.0785 (0.7940) acc1: 100.0000 (81.0410) acc5: 100.0000 (94.5881) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 3000/50000] eta: 0:03:45 loss: 0.8794 (0.8190) acc1: 100.0000 (80.5398) acc5: 100.0000 (94.3019) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 3100/50000] eta: 0:03:43 loss: 0.1602 (0.8417) acc1: 100.0000 (79.8130) acc5: 100.0000 (94.3567) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 3200/50000] eta: 0:03:41 loss: 0.0120 (0.8620) acc1: 100.0000 (79.2565) acc5: 100.0000 (94.0331) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 3300/50000] eta: 0:03:40 loss: 0.1979 (0.8830) acc1: 100.0000 (78.8549) acc5: 100.0000 (93.9109) time: 0.0041 data: 0.0003 max mem: 86
Test: [ 3400/50000] eta: 0:03:39 loss: 0.8700 (0.8987) acc1: 0.0000 (78.3593) acc5: 100.0000 (93.7959) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 3500/50000] eta: 0:03:38 loss: 0.0001 (0.9023) acc1: 100.0000 (77.8920) acc5: 100.0000 (93.7446) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 3600/50000] eta: 0:03:37 loss: 0.0002 (0.8921) acc1: 100.0000 (78.0894) acc5: 100.0000 (93.7795) time: 0.0048 data: 0.0002 max mem: 86
Test: [ 3700/50000] eta: 0:03:36 loss: 1.0290 (0.8984) acc1: 0.0000 (77.5736) acc5: 100.0000 (93.8665) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 3800/50000] eta: 0:03:35 loss: 0.0186 (0.9037) acc1: 100.0000 (77.2691) acc5: 100.0000 (93.9227) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 3900/50000] eta: 0:03:35 loss: 0.0721 (0.9016) acc1: 100.0000 (77.3135) acc5: 100.0000 (93.9503) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 4000/50000] eta: 0:03:34 loss: 0.0351 (0.9084) acc1: 100.0000 (77.3057) acc5: 100.0000 (93.8765) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 4100/50000] eta: 0:03:33 loss: 0.0029 (0.8997) acc1: 100.0000 (77.4689) acc5: 100.0000 (94.0015) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 4200/50000] eta: 0:03:32 loss: 0.0013 (0.8858) acc1: 100.0000 (77.7910) acc5: 100.0000 (94.1204) time: 0.0048 data: 0.0002 max mem: 86
Test: [ 4300/50000] eta: 0:03:32 loss: 0.0011 (0.8741) acc1: 100.0000 (78.1214) acc5: 100.0000 (94.1641) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 4400/50000] eta: 0:03:31 loss: 0.0000 (0.8666) acc1: 100.0000 (78.1641) acc5: 100.0000 (94.2513) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 4500/50000] eta: 0:03:30 loss: 0.0001 (0.8528) acc1: 100.0000 (78.5603) acc5: 100.0000 (94.3568) time: 0.0046 data: 0.0002 max mem: 86
Test: [ 4600/50000] eta: 0:03:30 loss: 0.0002 (0.8414) acc1: 100.0000 (78.8524) acc5: 100.0000 (94.3708) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 4700/50000] eta: 0:03:29 loss: 0.0031 (0.8295) acc1: 100.0000 (79.1534) acc5: 100.0000 (94.4905) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 4800/50000] eta: 0:03:29 loss: 0.0007 (0.8202) acc1: 100.0000 (79.4210) acc5: 100.0000 (94.5220) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 4900/50000] eta: 0:03:28 loss: 0.0011 (0.8136) acc1: 100.0000 (79.6164) acc5: 100.0000 (94.5725) time: 0.0047 data: 0.0002 max mem: 86
Test: [ 5000/50000] eta: 0:03:27 loss: 0.0025 (0.8080) acc1: 100.0000 (79.7041) acc5: 100.0000 (94.6411) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 5100/50000] eta: 0:03:26 loss: 0.3220 (0.8063) acc1: 100.0000 (79.7099) acc5: 100.0000 (94.6481) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 5200/50000] eta: 0:03:25 loss: 0.0062 (0.8012) acc1: 100.0000 (79.8500) acc5: 100.0000 (94.6164) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 5300/50000] eta: 0:03:25 loss: 0.0002 (0.7945) acc1: 100.0000 (80.0415) acc5: 100.0000 (94.6237) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 5400/50000] eta: 0:03:24 loss: 0.0015 (0.7994) acc1: 100.0000 (80.1889) acc5: 100.0000 (94.5751) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 5500/50000] eta: 0:03:24 loss: 0.0039 (0.8008) acc1: 100.0000 (80.2218) acc5: 100.0000 (94.5646) time: 0.0046 data: 0.0002 max mem: 86
Test: [ 5600/50000] eta: 0:03:23 loss: 0.0009 (0.7973) acc1: 100.0000 (80.3428) acc5: 100.0000 (94.5903) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 5700/50000] eta: 0:03:22 loss: 0.0045 (0.8016) acc1: 100.0000 (80.3368) acc5: 100.0000 (94.4922) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 5800/50000] eta: 0:03:21 loss: 0.0441 (0.8010) acc1: 100.0000 (80.4172) acc5: 100.0000 (94.4665) time: 0.0043 data: 0.0003 max mem: 86
Test: [ 5900/50000] eta: 0:03:20 loss: 0.0016 (0.8022) acc1: 100.0000 (80.5287) acc5: 100.0000 (94.4416) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 6000/50000] eta: 0:03:20 loss: 0.0303 (0.8104) acc1: 100.0000 (80.4199) acc5: 100.0000 (94.4176) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 6100/50000] eta: 0:03:19 loss: 0.0044 (0.8116) acc1: 100.0000 (80.3475) acc5: 100.0000 (94.4271) time: 0.0038 data: 0.0001 max mem: 86
Test: [ 6200/50000] eta: 0:03:18 loss: 0.0818 (0.8107) acc1: 100.0000 (80.3580) acc5: 100.0000 (94.3719) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 6300/50000] eta: 0:03:17 loss: 0.0063 (0.8139) acc1: 100.0000 (80.2888) acc5: 100.0000 (94.3977) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 6400/50000] eta: 0:03:17 loss: 0.0055 (0.8195) acc1: 100.0000 (80.3000) acc5: 100.0000 (94.2978) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 6500/50000] eta: 0:03:16 loss: 0.0001 (0.8117) acc1: 100.0000 (80.4645) acc5: 100.0000 (94.3393) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 6600/50000] eta: 0:03:15 loss: 0.0092 (0.8027) acc1: 100.0000 (80.6696) acc5: 100.0000 (94.4251) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 6700/50000] eta: 0:03:15 loss: 0.0012 (0.7978) acc1: 100.0000 (80.7790) acc5: 100.0000 (94.4337) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 6800/50000] eta: 0:03:14 loss: 0.0000 (0.7927) acc1: 100.0000 (80.8558) acc5: 100.0000 (94.4861) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 6900/50000] eta: 0:03:13 loss: 0.0122 (0.7864) acc1: 100.0000 (81.0172) acc5: 100.0000 (94.5225) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7000/50000] eta: 0:03:12 loss: 0.0003 (0.7786) acc1: 100.0000 (81.2170) acc5: 100.0000 (94.5579) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 7100/50000] eta: 0:03:11 loss: 0.0009 (0.7743) acc1: 100.0000 (81.3407) acc5: 100.0000 (94.5782) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7200/50000] eta: 0:03:11 loss: 0.0001 (0.7686) acc1: 100.0000 (81.4331) acc5: 100.0000 (94.6257) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 7300/50000] eta: 0:03:10 loss: 0.0002 (0.7626) acc1: 100.0000 (81.6053) acc5: 100.0000 (94.6583) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 7400/50000] eta: 0:03:09 loss: 0.0007 (0.7567) acc1: 100.0000 (81.7457) acc5: 100.0000 (94.7034) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7500/50000] eta: 0:03:10 loss: 0.0025 (0.7521) acc1: 100.0000 (81.8558) acc5: 100.0000 (94.7074) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7600/50000] eta: 0:03:09 loss: 0.0618 (0.7581) acc1: 100.0000 (81.7392) acc5: 100.0000 (94.6718) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7700/50000] eta: 0:03:08 loss: 0.2603 (0.7584) acc1: 100.0000 (81.7167) acc5: 100.0000 (94.6760) time: 0.0038 data: 0.0002 max mem: 86
Test: [ 7800/50000] eta: 0:03:08 loss: 0.3902 (0.7611) acc1: 100.0000 (81.6177) acc5: 100.0000 (94.6545) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 7900/50000] eta: 0:03:08 loss: 0.0121 (0.7564) acc1: 100.0000 (81.7745) acc5: 100.0000 (94.6969) time: 0.0047 data: 0.0002 max mem: 86
Test: [ 8000/50000] eta: 0:03:07 loss: 0.6661 (0.7616) acc1: 0.0000 (81.5523) acc5: 100.0000 (94.6882) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8100/50000] eta: 0:03:07 loss: 0.0326 (0.7601) acc1: 100.0000 (81.5578) acc5: 100.0000 (94.6797) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 8200/50000] eta: 0:03:07 loss: 0.6533 (0.7703) acc1: 100.0000 (81.4413) acc5: 100.0000 (94.5982) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8300/50000] eta: 0:03:06 loss: 0.5622 (0.7768) acc1: 100.0000 (81.3637) acc5: 100.0000 (94.5428) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8400/50000] eta: 0:03:06 loss: 0.9667 (0.7885) acc1: 0.0000 (80.9666) acc5: 100.0000 (94.4768) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 8500/50000] eta: 0:03:06 loss: 0.0256 (0.7869) acc1: 100.0000 (81.0022) acc5: 100.0000 (94.5065) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8600/50000] eta: 0:03:05 loss: 0.0389 (0.7950) acc1: 100.0000 (80.8394) acc5: 100.0000 (94.4774) time: 0.0043 data: 0.0002 max mem: 86
Test: [ 8700/50000] eta: 0:03:05 loss: 0.2748 (0.7973) acc1: 100.0000 (80.7378) acc5: 100.0000 (94.4604) time: 0.0045 data: 0.0002 max mem: 86
Test: [ 8800/50000] eta: 0:03:04 loss: 0.1653 (0.8007) acc1: 100.0000 (80.7181) acc5: 100.0000 (94.4438) time: 0.0046 data: 0.0003 max mem: 86
Test: [ 8900/50000] eta: 0:03:04 loss: 0.1470 (0.7986) acc1: 100.0000 (80.7550) acc5: 100.0000 (94.4388) time: 0.0046 data: 0.0002 max mem: 86
Test: [ 9000/50000] eta: 0:03:03 loss: 0.2561 (0.7993) acc1: 100.0000 (80.7133) acc5: 100.0000 (94.4340) time: 0.0042 data: 0.0002 max mem: 86
Test: [ 9100/50000] eta: 0:03:03 loss: 0.0316 (0.7989) acc1: 100.0000 (80.6725) acc5: 100.0000 (94.4621) time: 0.0036 data: 0.0002 max mem: 86
Test: [ 9200/50000] eta: 0:03:02 loss: 0.3239 (0.7978) acc1: 100.0000 (80.6543) acc5: 100.0000 (94.4680) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 9300/50000] eta: 0:03:01 loss: 0.1094 (0.8015) acc1: 100.0000 (80.5075) acc5: 100.0000 (94.4737) time: 0.0039 data: 0.0002 max mem: 86
Test: [ 9400/50000] eta: 0:03:00 loss: 0.2506 (0.8040) acc1: 100.0000 (80.4489) acc5: 100.0000 (94.4580) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 9500/50000] eta: 0:03:00 loss: 0.1578 (0.8035) acc1: 100.0000 (80.4021) acc5: 100.0000 (94.4743) time: 0.0047 data: 0.0002 max mem: 86
Test: [ 9600/50000] eta: 0:03:00 loss: 0.0188 (0.8056) acc1: 100.0000 (80.4187) acc5: 100.0000 (94.4381) time: 0.0044 data: 0.0002 max mem: 86
Test: [ 9700/50000] eta: 0:02:59 loss: 0.4514 (0.8102) acc1: 100.0000 (80.2495) acc5: 100.0000 (94.4233) time: 0.0041 data: 0.0002 max mem: 86
Test: [ 9800/50000] eta: 0:02:58 loss: 0.0612 (0.8095) acc1: 100.0000 (80.2775) acc5: 100.0000 (94.4189) time: 0.0040 data: 0.0002 max mem: 86
Test: [ 9900/50000] eta: 0:02:58 loss: 0.3089 (0.8120) acc1: 100.0000 (80.2242) acc5: 100.0000 (94.4248) time: 0.0043 data: 0.0002 max mem: 86
Test: [10000/50000] eta: 0:02:57 loss: 0.0415 (0.8116) acc1: 100.0000 (80.2320) acc5: 100.0000 (94.4606) time: 0.0041 data: 0.0002 max mem: 86
Test: [10100/50000] eta: 0:02:57 loss: 0.5171 (0.8145) acc1: 100.0000 (80.0614) acc5: 100.0000 (94.4758) time: 0.0040 data: 0.0002 max mem: 86
Test: [10200/50000] eta: 0:02:56 loss: 0.0181 (0.8114) acc1: 100.0000 (80.0902) acc5: 100.0000 (94.5103) time: 0.0041 data: 0.0002 max mem: 86
Test: [10300/50000] eta: 0:02:56 loss: 0.0997 (0.8119) acc1: 100.0000 (80.0408) acc5: 100.0000 (94.5345) time: 0.0041 data: 0.0002 max mem: 86
Test: [10400/50000] eta: 0:02:55 loss: 0.0290 (0.8098) acc1: 100.0000 (80.0788) acc5: 100.0000 (94.5582) time: 0.0041 data: 0.0002 max mem: 86
Test: [10500/50000] eta: 0:02:55 loss: 0.0330 (0.8097) acc1: 100.0000 (80.1257) acc5: 100.0000 (94.5624) time: 0.0041 data: 0.0002 max mem: 86
Test: [10600/50000] eta: 0:02:54 loss: 0.1191 (0.8095) acc1: 100.0000 (80.1057) acc5: 100.0000 (94.5477) time: 0.0042 data: 0.0002 max mem: 86
Test: [10700/50000] eta: 0:02:54 loss: 0.0099 (0.8100) acc1: 100.0000 (80.0953) acc5: 100.0000 (94.5426) time: 0.0042 data: 0.0002 max mem: 86
Test: [10800/50000] eta: 0:02:53 loss: 0.2072 (0.8087) acc1: 100.0000 (80.1315) acc5: 100.0000 (94.5468) time: 0.0043 data: 0.0002 max mem: 86
Test: [10900/50000] eta: 0:02:53 loss: 0.0128 (0.8042) acc1: 100.0000 (80.2312) acc5: 100.0000 (94.5785) time: 0.0047 data: 0.0002 max mem: 86
Test: [11000/50000] eta: 0:02:52 loss: 0.0486 (0.8027) acc1: 100.0000 (80.2382) acc5: 100.0000 (94.6187) time: 0.0042 data: 0.0002 max mem: 86
Test: [11100/50000] eta: 0:02:52 loss: 0.0621 (0.8054) acc1: 100.0000 (80.1910) acc5: 100.0000 (94.6041) time: 0.0042 data: 0.0002 max mem: 86
Test: [11200/50000] eta: 0:02:51 loss: 0.0130 (0.8056) acc1: 100.0000 (80.1893) acc5: 100.0000 (94.5719) time: 0.0042 data: 0.0002 max mem: 86
Test: [11300/50000] eta: 0:02:51 loss: 0.0220 (0.8028) acc1: 100.0000 (80.2407) acc5: 100.0000 (94.5934) time: 0.0042 data: 0.0002 max mem: 86
Test: [11400/50000] eta: 0:02:50 loss: 0.0704 (0.8124) acc1: 100.0000 (80.1333) acc5: 100.0000 (94.5180) time: 0.0042 data: 0.0002 max mem: 86
Test: [11500/50000] eta: 0:02:50 loss: 0.0229 (0.8090) acc1: 100.0000 (80.2278) acc5: 100.0000 (94.5135) time: 0.0043 data: 0.0002 max mem: 86
Test: [11600/50000] eta: 0:02:49 loss: 0.3336 (0.8100) acc1: 100.0000 (80.1396) acc5: 100.0000 (94.5005) time: 0.0044 data: 0.0002 max mem: 86
Test: [11700/50000] eta: 0:02:49 loss: 0.0793 (0.8110) acc1: 100.0000 (80.1214) acc5: 100.0000 (94.4877) time: 0.0042 data: 0.0002 max mem: 86
Test: [11800/50000] eta: 0:02:48 loss: 0.0249 (0.8095) acc1: 100.0000 (80.1542) acc5: 100.0000 (94.5174) time: 0.0040 data: 0.0002 max mem: 86
Test: [11900/50000] eta: 0:02:47 loss: 0.0822 (0.8103) acc1: 100.0000 (80.1361) acc5: 100.0000 (94.4711) time: 0.0042 data: 0.0002 max mem: 86
Test: [12000/50000] eta: 0:02:47 loss: 0.0385 (0.8100) acc1: 100.0000 (80.1350) acc5: 100.0000 (94.5088) time: 0.0041 data: 0.0002 max mem: 86
Test: [12100/50000] eta: 0:02:46 loss: 0.2712 (0.8136) acc1: 100.0000 (79.9273) acc5: 100.0000 (94.5376) time: 0.0041 data: 0.0002 max mem: 86
Test: [12200/50000] eta: 0:02:46 loss: 0.0601 (0.8135) acc1: 100.0000 (79.8951) acc5: 100.0000 (94.5496) time: 0.0040 data: 0.0002 max mem: 86
Test: [12300/50000] eta: 0:02:45 loss: 0.0080 (0.8108) acc1: 100.0000 (79.9610) acc5: 100.0000 (94.5695) time: 0.0043 data: 0.0002 max mem: 86
Test: [12400/50000] eta: 0:02:45 loss: 0.0194 (0.8125) acc1: 100.0000 (79.9129) acc5: 100.0000 (94.5488) time: 0.0044 data: 0.0002 max mem: 86
Test: [12500/50000] eta: 0:02:44 loss: 0.2914 (0.8163) acc1: 100.0000 (79.7696) acc5: 100.0000 (94.5444) time: 0.0042 data: 0.0002 max mem: 86
Test: [12600/50000] eta: 0:02:44 loss: 0.0031 (0.8145) acc1: 100.0000 (79.7794) acc5: 100.0000 (94.5877) time: 0.0040 data: 0.0002 max mem: 86
Test: [12700/50000] eta: 0:02:43 loss: 0.0760 (0.8128) acc1: 100.0000 (79.7969) acc5: 100.0000 (94.6146) time: 0.0039 data: 0.0002 max mem: 86
Test: [12800/50000] eta: 0:02:43 loss: 0.0001 (0.8072) acc1: 100.0000 (79.9234) acc5: 100.0000 (94.6489) time: 0.0038 data: 0.0002 max mem: 86
Test: [12900/50000] eta: 0:02:42 loss: 0.0351 (0.8071) acc1: 100.0000 (79.9008) acc5: 100.0000 (94.6516) time: 0.0040 data: 0.0002 max mem: 86
Test: [13000/50000] eta: 0:02:42 loss: 0.0030 (0.8019) acc1: 100.0000 (80.0323) acc5: 100.0000 (94.6927) time: 0.0040 data: 0.0002 max mem: 86
Test: [13100/50000] eta: 0:02:41 loss: 0.0005 (0.7990) acc1: 100.0000 (80.1084) acc5: 100.0000 (94.7180) time: 0.0039 data: 0.0002 max mem: 86
Test: [13200/50000] eta: 0:02:40 loss: 0.0728 (0.7972) acc1: 100.0000 (80.1454) acc5: 100.0000 (94.7352) time: 0.0037 data: 0.0002 max mem: 86
Test: [13300/50000] eta: 0:02:40 loss: 0.6036 (0.7994) acc1: 100.0000 (80.0241) acc5: 100.0000 (94.7372) time: 0.0037 data: 0.0002 max mem: 86
Test: [13400/50000] eta: 0:02:39 loss: 0.0842 (0.8006) acc1: 100.0000 (79.9269) acc5: 100.0000 (94.7392) time: 0.0050 data: 0.0002 max mem: 86
Test: [13500/50000] eta: 0:02:39 loss: 0.3301 (0.7995) acc1: 100.0000 (79.9348) acc5: 100.0000 (94.7634) time: 0.0039 data: 0.0002 max mem: 86
Test: [13600/50000] eta: 0:02:38 loss: 0.4816 (0.8014) acc1: 100.0000 (79.8765) acc5: 100.0000 (94.7651) time: 0.0039 data: 0.0001 max mem: 86
Test: [13700/50000] eta: 0:02:38 loss: 0.0488 (0.8076) acc1: 100.0000 (79.8044) acc5: 100.0000 (94.7376) time: 0.0042 data: 0.0002 max mem: 86
Test: [13800/50000] eta: 0:02:37 loss: 0.0009 (0.8044) acc1: 100.0000 (79.9145) acc5: 100.0000 (94.7685) time: 0.0037 data: 0.0002 max mem: 86
Test: [13900/50000] eta: 0:02:37 loss: 0.2700 (0.8054) acc1: 100.0000 (79.8576) acc5: 100.0000 (94.7917) time: 0.0040 data: 0.0002 max mem: 86
Test: [14000/50000] eta: 0:02:36 loss: 0.0046 (0.8038) acc1: 100.0000 (79.8657) acc5: 100.0000 (94.8218) time: 0.0038 data: 0.0002 max mem: 86
Test: [14100/50000] eta: 0:02:36 loss: 0.6475 (0.8075) acc1: 100.0000 (79.7603) acc5: 100.0000 (94.8089) time: 0.0049 data: 0.0002 max mem: 86
Test: [14200/50000] eta: 0:02:35 loss: 0.0036 (0.8107) acc1: 100.0000 (79.5930) acc5: 100.0000 (94.8032) time: 0.0036 data: 0.0002 max mem: 86
Test: [14300/50000] eta: 0:02:34 loss: 0.8769 (0.8150) acc1: 0.0000 (79.5189) acc5: 100.0000 (94.7416) time: 0.0038 data: 0.0003 max mem: 86
Test: [14400/50000] eta: 0:02:34 loss: 0.0047 (0.8161) acc1: 100.0000 (79.5500) acc5: 100.0000 (94.7365) time: 0.0037 data: 0.0002 max mem: 86
Test: [14500/50000] eta: 0:02:33 loss: 0.0571 (0.8135) acc1: 100.0000 (79.5945) acc5: 100.0000 (94.7659) time: 0.0038 data: 0.0002 max mem: 86
Test: [14600/50000] eta: 0:02:33 loss: 0.0004 (0.8125) acc1: 100.0000 (79.6384) acc5: 100.0000 (94.7743) time: 0.0035 data: 0.0002 max mem: 86
Test: [14700/50000] eta: 0:02:32 loss: 0.0237 (0.8089) acc1: 100.0000 (79.7293) acc5: 100.0000 (94.8031) time: 0.0099 data: 0.0060 max mem: 86
Test: [14800/50000] eta: 0:02:32 loss: 0.0363 (0.8052) acc1: 100.0000 (79.7784) acc5: 100.0000 (94.8382) time: 0.0038 data: 0.0002 max mem: 86
Test: [14900/50000] eta: 0:02:31 loss: 0.0038 (0.8080) acc1: 100.0000 (79.7933) acc5: 100.0000 (94.8259) time: 0.0041 data: 0.0002 max mem: 86
Test: [15000/50000] eta: 0:02:31 loss: 0.1103 (0.8090) acc1: 100.0000 (79.7680) acc5: 100.0000 (94.8203) time: 0.0050 data: 0.0002 max mem: 86
Test: [15100/50000] eta: 0:02:31 loss: 0.0379 (0.8061) acc1: 100.0000 (79.8027) acc5: 100.0000 (94.8414) time: 0.0043 data: 0.0002 max mem: 86
Test: [15200/50000] eta: 0:02:30 loss: 0.2943 (0.8078) acc1: 100.0000 (79.7448) acc5: 100.0000 (94.8622) time: 0.0042 data: 0.0002 max mem: 86
Test: [15300/50000] eta: 0:02:30 loss: 0.0289 (0.8078) acc1: 100.0000 (79.7137) acc5: 100.0000 (94.8827) time: 0.0045 data: 0.0002 max mem: 86
Test: [15400/50000] eta: 0:02:29 loss: 0.0153 (0.8058) acc1: 100.0000 (79.7546) acc5: 100.0000 (94.8964) time: 0.0042 data: 0.0002 max mem: 86
Test: [15500/50000] eta: 0:02:29 loss: 0.0358 (0.8057) acc1: 100.0000 (79.7239) acc5: 100.0000 (94.8907) time: 0.0038 data: 0.0002 max mem: 86
Test: [15600/50000] eta: 0:02:28 loss: 0.1787 (0.8104) acc1: 100.0000 (79.6103) acc5: 100.0000 (94.8529) time: 0.0039 data: 0.0002 max mem: 86
Test: [15700/50000] eta: 0:02:28 loss: 0.2585 (0.8120) acc1: 100.0000 (79.5172) acc5: 100.0000 (94.8602) time: 0.0040 data: 0.0002 max mem: 86
Test: [15800/50000] eta: 0:02:27 loss: 0.0670 (0.8148) acc1: 100.0000 (79.5140) acc5: 100.0000 (94.8168) time: 0.0038 data: 0.0002 max mem: 86
Test: [15900/50000] eta: 0:02:27 loss: 0.0272 (0.8127) acc1: 100.0000 (79.5736) acc5: 100.0000 (94.8368) time: 0.0039 data: 0.0002 max mem: 86
Test: [16000/50000] eta: 0:02:26 loss: 0.0547 (0.8122) acc1: 100.0000 (79.6013) acc5: 100.0000 (94.8503) time: 0.0041 data: 0.0002 max mem: 86
Test: [16100/50000] eta: 0:02:26 loss: 0.0000 (0.8089) acc1: 100.0000 (79.6907) acc5: 100.0000 (94.8699) time: 0.0041 data: 0.0002 max mem: 86
Test: [16200/50000] eta: 0:02:25 loss: 0.0006 (0.8054) acc1: 100.0000 (79.7852) acc5: 100.0000 (94.8954) time: 0.0040 data: 0.0002 max mem: 86
Test: [16300/50000] eta: 0:02:25 loss: 0.0003 (0.8019) acc1: 100.0000 (79.8724) acc5: 100.0000 (94.9267) time: 0.0039 data: 0.0002 max mem: 86
Test: [16400/50000] eta: 0:02:24 loss: 0.0062 (0.8004) acc1: 100.0000 (79.9281) acc5: 100.0000 (94.9332) time: 0.0041 data: 0.0002 max mem: 86
Test: [16500/50000] eta: 0:02:24 loss: 0.0020 (0.8020) acc1: 100.0000 (79.9224) acc5: 100.0000 (94.8973) time: 0.0039 data: 0.0002 max mem: 86
Test: [16600/50000] eta: 0:02:24 loss: 0.0321 (0.8014) acc1: 100.0000 (79.9108) acc5: 100.0000 (94.9160) time: 0.0043 data: 0.0002 max mem: 86
Test: [16700/50000] eta: 0:02:23 loss: 0.0001 (0.7972) acc1: 100.0000 (80.0132) acc5: 100.0000 (94.9464) time: 0.0043 data: 0.0002 max mem: 86
Test: [16800/50000] eta: 0:02:23 loss: 0.0031 (0.7957) acc1: 100.0000 (80.0488) acc5: 100.0000 (94.9646) time: 0.0044 data: 0.0002 max mem: 86
Test: [16900/50000] eta: 0:02:22 loss: 0.0138 (0.7970) acc1: 100.0000 (80.0781) acc5: 100.0000 (94.9707) time: 0.0043 data: 0.0002 max mem: 86
Test: [17000/50000] eta: 0:02:22 loss: 0.0021 (0.7951) acc1: 100.0000 (80.1306) acc5: 100.0000 (94.9768) time: 0.0040 data: 0.0002 max mem: 86
Test: [17100/50000] eta: 0:02:21 loss: 1.9055 (0.7958) acc1: 0.0000 (80.1181) acc5: 100.0000 (94.9594) time: 0.0038 data: 0.0002 max mem: 86
Test: [17200/50000] eta: 0:02:21 loss: 0.0138 (0.7940) acc1: 100.0000 (80.1523) acc5: 100.0000 (94.9770) time: 0.0041 data: 0.0002 max mem: 86
Test: [17300/50000] eta: 0:02:20 loss: 0.4505 (0.7939) acc1: 100.0000 (80.1225) acc5: 100.0000 (94.9829) time: 0.0044 data: 0.0002 max mem: 86
Test: [17400/50000] eta: 0:02:20 loss: 0.0002 (0.7931) acc1: 100.0000 (80.1448) acc5: 100.0000 (94.9601) time: 0.0038 data: 0.0002 max mem: 86
Test: [17500/50000] eta: 0:02:19 loss: 0.3345 (0.7926) acc1: 100.0000 (80.0640) acc5: 100.0000 (94.9774) time: 0.0039 data: 0.0002 max mem: 86
Test: [17600/50000] eta: 0:02:19 loss: 0.0260 (0.7892) acc1: 100.0000 (80.1432) acc5: 100.0000 (95.0003) time: 0.0039 data: 0.0003 max mem: 86
Test: [17700/50000] eta: 0:02:18 loss: 0.0452 (0.7887) acc1: 100.0000 (80.1198) acc5: 100.0000 (95.0172) time: 0.0038 data: 0.0002 max mem: 86
Test: [17800/50000] eta: 0:02:18 loss: 0.0026 (0.7865) acc1: 100.0000 (80.1865) acc5: 100.0000 (95.0171) time: 0.0037 data: 0.0002 max mem: 86
Test: [17900/50000] eta: 0:02:17 loss: 0.0605 (0.7897) acc1: 100.0000 (80.1017) acc5: 100.0000 (95.0226) time: 0.0037 data: 0.0002 max mem: 86
Test: [18000/50000] eta: 0:02:17 loss: 0.1980 (0.7939) acc1: 100.0000 (79.9956) acc5: 100.0000 (95.0114) time: 0.0038 data: 0.0002 max mem: 86
Test: [18100/50000] eta: 0:02:17 loss: 0.0005 (0.7933) acc1: 100.0000 (79.9845) acc5: 100.0000 (95.0113) time: 0.0039 data: 0.0002 max mem: 86
Test: [18200/50000] eta: 0:02:16 loss: 0.0000 (0.7923) acc1: 100.0000 (80.0231) acc5: 100.0000 (95.0113) time: 0.0042 data: 0.0002 max mem: 86
Test: [18300/50000] eta: 0:02:16 loss: 0.0105 (0.7899) acc1: 100.0000 (80.1049) acc5: 100.0000 (95.0221) time: 0.0041 data: 0.0002 max mem: 86
Test: [18400/50000] eta: 0:02:15 loss: 0.0868 (0.7898) acc1: 100.0000 (80.1261) acc5: 100.0000 (95.0166) time: 0.0037 data: 0.0002 max mem: 86
Test: [18500/50000] eta: 0:02:15 loss: 0.0747 (0.7911) acc1: 100.0000 (80.1146) acc5: 100.0000 (94.9949) time: 0.0037 data: 0.0002 max mem: 86
Test: [18600/50000] eta: 0:02:14 loss: 0.0352 (0.7937) acc1: 100.0000 (80.1086) acc5: 100.0000 (94.9734) time: 0.0037 data: 0.0002 max mem: 86
Test: [18700/50000] eta: 0:02:14 loss: 0.0732 (0.7940) acc1: 100.0000 (80.1294) acc5: 100.0000 (94.9575) time: 0.0039 data: 0.0002 max mem: 86
Test: [18800/50000] eta: 0:02:13 loss: 0.0474 (0.7955) acc1: 100.0000 (80.1074) acc5: 100.0000 (94.9418) time: 0.0037 data: 0.0002 max mem: 86
Test: [18900/50000] eta: 0:02:13 loss: 0.1086 (0.7941) acc1: 100.0000 (80.1386) acc5: 100.0000 (94.9474) time: 0.0038 data: 0.0002 max mem: 86
Test: [19000/50000] eta: 0:02:12 loss: 0.0733 (0.7935) acc1: 100.0000 (80.1484) acc5: 100.0000 (94.9529) time: 0.0039 data: 0.0002 max mem: 86
Test: [19100/50000] eta: 0:02:12 loss: 1.5252 (0.7983) acc1: 0.0000 (80.0063) acc5: 100.0000 (94.9322) time: 0.0037 data: 0.0002 max mem: 86
Test: [19200/50000] eta: 0:02:11 loss: 0.0863 (0.8004) acc1: 100.0000 (79.9385) acc5: 100.0000 (94.9273) time: 0.0041 data: 0.0002 max mem: 86
Test: [19300/50000] eta: 0:02:11 loss: 0.7411 (0.7994) acc1: 100.0000 (79.9285) acc5: 100.0000 (94.9277) time: 0.0042 data: 0.0002 max mem: 86
Test: [19400/50000] eta: 0:02:10 loss: 0.0001 (0.8005) acc1: 100.0000 (79.9083) acc5: 100.0000 (94.9281) time: 0.0042 data: 0.0002 max mem: 86
Test: [19500/50000] eta: 0:02:10 loss: 0.0033 (0.8001) acc1: 100.0000 (79.9190) acc5: 100.0000 (94.9182) time: 0.0037 data: 0.0002 max mem: 86
Test: [19600/50000] eta: 0:02:09 loss: 0.0073 (0.8024) acc1: 100.0000 (79.8786) acc5: 100.0000 (94.9033) time: 0.0038 data: 0.0002 max mem: 86
Test: [19700/50000] eta: 0:02:09 loss: 0.0179 (0.8002) acc1: 100.0000 (79.9452) acc5: 100.0000 (94.9140) time: 0.0038 data: 0.0002 max mem: 86
Test: [19800/50000] eta: 0:02:08 loss: 0.1315 (0.8002) acc1: 100.0000 (79.9303) acc5: 100.0000 (94.9043) time: 0.0038 data: 0.0002 max mem: 86
Test: [19900/50000] eta: 0:02:08 loss: 0.0081 (0.7988) acc1: 100.0000 (79.9910) acc5: 100.0000 (94.9148) time: 0.0037 data: 0.0002 max mem: 86
Test: [20000/50000] eta: 0:02:07 loss: 0.1578 (0.7995) acc1: 100.0000 (79.9760) acc5: 100.0000 (94.9053) time: 0.0036 data: 0.0002 max mem: 86
Test: [20100/50000] eta: 0:02:07 loss: 0.1097 (0.8019) acc1: 100.0000 (79.9015) acc5: 100.0000 (94.8858) time: 0.0040 data: 0.0002 max mem: 86
Test: [20200/50000] eta: 0:02:06 loss: 0.0078 (0.8048) acc1: 100.0000 (79.8673) acc5: 100.0000 (94.8567) time: 0.0039 data: 0.0002 max mem: 86
Test: [20300/50000] eta: 0:02:06 loss: 0.0131 (0.8054) acc1: 100.0000 (79.8778) acc5: 100.0000 (94.8278) time: 0.0039 data: 0.0002 max mem: 86
Test: [20400/50000] eta: 0:02:05 loss: 0.0012 (0.8049) acc1: 100.0000 (79.8882) acc5: 100.0000 (94.8434) time: 0.0036 data: 0.0002 max mem: 86
Test: [20500/50000] eta: 0:02:05 loss: 0.3294 (0.8079) acc1: 100.0000 (79.8303) acc5: 100.0000 (94.8490) time: 0.0038 data: 0.0002 max mem: 86
Test: [20600/50000] eta: 0:02:04 loss: 0.0362 (0.8077) acc1: 100.0000 (79.8262) acc5: 100.0000 (94.8498) time: 0.0039 data: 0.0002 max mem: 86
Test: [20700/50000] eta: 0:02:04 loss: 0.7428 (0.8134) acc1: 0.0000 (79.7111) acc5: 100.0000 (94.8022) time: 0.0041 data: 0.0002 max mem: 86
Test: [20800/50000] eta: 0:02:03 loss: 1.3721 (0.8242) acc1: 0.0000 (79.5058) acc5: 100.0000 (94.6685) time: 0.0036 data: 0.0002 max mem: 86
Test: [20900/50000] eta: 0:02:03 loss: 0.0012 (0.8234) acc1: 100.0000 (79.5225) acc5: 100.0000 (94.6701) time: 0.0039 data: 0.0002 max mem: 86
Test: [21000/50000] eta: 0:02:02 loss: 0.2224 (0.8257) acc1: 100.0000 (79.4724) acc5: 100.0000 (94.6574) time: 0.0038 data: 0.0002 max mem: 86
Test: [21100/50000] eta: 0:02:02 loss: 0.2699 (0.8258) acc1: 100.0000 (79.4749) acc5: 100.0000 (94.6590) time: 0.0039 data: 0.0002 max mem: 86
Test: [21200/50000] eta: 0:02:02 loss: 0.2430 (0.8297) acc1: 100.0000 (79.3925) acc5: 100.0000 (94.6465) time: 0.0039 data: 0.0002 max mem: 86
Test: [21300/50000] eta: 0:02:01 loss: 0.0005 (0.8302) acc1: 100.0000 (79.3953) acc5: 100.0000 (94.6387) time: 0.0038 data: 0.0002 max mem: 86
Test: [21400/50000] eta: 0:02:01 loss: 0.1168 (0.8315) acc1: 100.0000 (79.3748) acc5: 100.0000 (94.6358) time: 0.0039 data: 0.0002 max mem: 86
Test: [21500/50000] eta: 0:02:00 loss: 0.0134 (0.8331) acc1: 100.0000 (79.3591) acc5: 100.0000 (94.6096) time: 0.0040 data: 0.0002 max mem: 86
Test: [21600/50000] eta: 0:02:00 loss: 0.1121 (0.8323) acc1: 100.0000 (79.3713) acc5: 100.0000 (94.6067) time: 0.0038 data: 0.0002 max mem: 86
Test: [21700/50000] eta: 0:01:59 loss: 0.0110 (0.8337) acc1: 100.0000 (79.3374) acc5: 100.0000 (94.6039) time: 0.0040 data: 0.0003 max mem: 86
Test: [21800/50000] eta: 0:01:59 loss: 0.7444 (0.8395) acc1: 0.0000 (79.1982) acc5: 100.0000 (94.5369) time: 0.0038 data: 0.0002 max mem: 86
Test: [21900/50000] eta: 0:01:58 loss: 0.2301 (0.8426) acc1: 100.0000 (79.1425) acc5: 100.0000 (94.5071) time: 0.0037 data: 0.0002 max mem: 86
Test: [22000/50000] eta: 0:01:58 loss: 0.0000 (0.8459) acc1: 100.0000 (79.0964) acc5: 100.0000 (94.4548) time: 0.0038 data: 0.0002 max mem: 86
Test: [22100/50000] eta: 0:01:57 loss: 0.1825 (0.8474) acc1: 100.0000 (79.0733) acc5: 100.0000 (94.4392) time: 0.0036 data: 0.0002 max mem: 86
Test: [22200/50000] eta: 0:01:57 loss: 0.0235 (0.8493) acc1: 100.0000 (79.0370) acc5: 100.0000 (94.4327) time: 0.0037 data: 0.0002 max mem: 86
Test: [22300/50000] eta: 0:01:56 loss: 1.0514 (0.8507) acc1: 0.0000 (79.0054) acc5: 100.0000 (94.4173) time: 0.0036 data: 0.0002 max mem: 86
Test: [22400/50000] eta: 0:01:56 loss: 0.0888 (0.8552) acc1: 100.0000 (78.9072) acc5: 100.0000 (94.3753) time: 0.0036 data: 0.0002 max mem: 86
Test: [22500/50000] eta: 0:01:55 loss: 0.0557 (0.8541) acc1: 100.0000 (78.9298) acc5: 100.0000 (94.3825) time: 0.0045 data: 0.0003 max mem: 86
Test: [22600/50000] eta: 0:01:55 loss: 0.0050 (0.8534) acc1: 100.0000 (78.9567) acc5: 100.0000 (94.3719) time: 0.0037 data: 0.0002 max mem: 86
Test: [22700/50000] eta: 0:01:54 loss: 0.2502 (0.8564) acc1: 100.0000 (78.8952) acc5: 100.0000 (94.3439) time: 0.0038 data: 0.0002 max mem: 86
Test: [22800/50000] eta: 0:01:54 loss: 0.0013 (0.8606) acc1: 100.0000 (78.7948) acc5: 100.0000 (94.2766) time: 0.0039 data: 0.0002 max mem: 86
Test: [22900/50000] eta: 0:01:53 loss: 0.1173 (0.8620) acc1: 100.0000 (78.7608) acc5: 100.0000 (94.2492) time: 0.0037 data: 0.0002 max mem: 86
Test: [23000/50000] eta: 0:01:53 loss: 0.3043 (0.8637) acc1: 100.0000 (78.7314) acc5: 100.0000 (94.2263) time: 0.0038 data: 0.0002 max mem: 86
Test: [23100/50000] eta: 0:01:52 loss: 1.0789 (0.8663) acc1: 0.0000 (78.6156) acc5: 100.0000 (94.2080) time: 0.0038 data: 0.0002 max mem: 86
Test: [23200/50000] eta: 0:01:52 loss: 0.4022 (0.8728) acc1: 100.0000 (78.5311) acc5: 100.0000 (94.1554) time: 0.0037 data: 0.0002 max mem: 86
Test: [23300/50000] eta: 0:01:52 loss: 1.2780 (0.8800) acc1: 0.0000 (78.4044) acc5: 100.0000 (94.0346) time: 0.0053 data: 0.0013 max mem: 86
Test: [23400/50000] eta: 0:01:51 loss: 0.0102 (0.8784) acc1: 100.0000 (78.4411) acc5: 100.0000 (94.0473) time: 0.0041 data: 0.0002 max mem: 86
Test: [23500/50000] eta: 0:01:51 loss: 1.7649 (0.8824) acc1: 0.0000 (78.3584) acc5: 100.0000 (94.0088) time: 0.0040 data: 0.0002 max mem: 86
Test: [23600/50000] eta: 0:01:50 loss: 0.0051 (0.8853) acc1: 100.0000 (78.3314) acc5: 100.0000 (93.9791) time: 0.0043 data: 0.0002 max mem: 86
Test: [23700/50000] eta: 0:01:50 loss: 0.2883 (0.8852) acc1: 100.0000 (78.3511) acc5: 100.0000 (93.9707) time: 0.0040 data: 0.0002 max mem: 86
Test: [23800/50000] eta: 0:01:49 loss: 0.0004 (0.8854) acc1: 100.0000 (78.3581) acc5: 100.0000 (93.9582) time: 0.0041 data: 0.0002 max mem: 86
Test: [23900/50000] eta: 0:01:49 loss: 0.0361 (0.8858) acc1: 100.0000 (78.3566) acc5: 100.0000 (93.9417) time: 0.0040 data: 0.0002 max mem: 86
Test: [24000/50000] eta: 0:01:49 loss: 1.5626 (0.8937) acc1: 0.0000 (78.2051) acc5: 100.0000 (93.8628) time: 0.0042 data: 0.0002 max mem: 86
Test: [24100/50000] eta: 0:01:48 loss: 0.1333 (0.8958) acc1: 100.0000 (78.1503) acc5: 100.0000 (93.8467) time: 0.0038 data: 0.0002 max mem: 86
Test: [24200/50000] eta: 0:01:48 loss: 0.0076 (0.8986) acc1: 100.0000 (78.0340) acc5: 100.0000 (93.8391) time: 0.0040 data: 0.0002 max mem: 86
Test: [24300/50000] eta: 0:01:47 loss: 1.0008 (0.9013) acc1: 0.0000 (77.9433) acc5: 100.0000 (93.8315) time: 0.0039 data: 0.0002 max mem: 86
Test: [24400/50000] eta: 0:01:47 loss: 0.1565 (0.9017) acc1: 100.0000 (77.9394) acc5: 100.0000 (93.8281) time: 0.0041 data: 0.0002 max mem: 86
Test: [24500/50000] eta: 0:01:46 loss: 0.0557 (0.9064) acc1: 100.0000 (77.8499) acc5: 100.0000 (93.7839) time: 0.0044 data: 0.0003 max mem: 86
Test: [24600/50000] eta: 0:01:46 loss: 0.0539 (0.9073) acc1: 100.0000 (77.8383) acc5: 100.0000 (93.7889) time: 0.0047 data: 0.0002 max mem: 86
Test: [24700/50000] eta: 0:01:46 loss: 2.4288 (0.9127) acc1: 0.0000 (77.7337) acc5: 100.0000 (93.7250) time: 0.0041 data: 0.0002 max mem: 86
Test: [24800/50000] eta: 0:01:45 loss: 0.0067 (0.9131) acc1: 100.0000 (77.7146) acc5: 100.0000 (93.7261) time: 0.0041 data: 0.0002 max mem: 86
Test: [24900/50000] eta: 0:01:45 loss: 0.3004 (0.9131) acc1: 100.0000 (77.7037) acc5: 100.0000 (93.7312) time: 0.0041 data: 0.0002 max mem: 86
Test: [25000/50000] eta: 0:01:44 loss: 1.6132 (0.9186) acc1: 0.0000 (77.5929) acc5: 100.0000 (93.6643) time: 0.0042 data: 0.0002 max mem: 86
Test: [25100/50000] eta: 0:01:44 loss: 1.4236 (0.9222) acc1: 0.0000 (77.5427) acc5: 100.0000 (93.6178) time: 0.0042 data: 0.0002 max mem: 86
Test: [25200/50000] eta: 0:01:44 loss: 0.2022 (0.9257) acc1: 100.0000 (77.4811) acc5: 100.0000 (93.5637) time: 0.0040 data: 0.0002 max mem: 86
Test: [25300/50000] eta: 0:01:43 loss: 0.8040 (0.9300) acc1: 0.0000 (77.3962) acc5: 100.0000 (93.5220) time: 0.0042 data: 0.0002 max mem: 86
Test: [25400/50000] eta: 0:01:43 loss: 0.0320 (0.9325) acc1: 100.0000 (77.3670) acc5: 100.0000 (93.5042) time: 0.0040 data: 0.0002 max mem: 86
Test: [25500/50000] eta: 0:01:42 loss: 0.4818 (0.9350) acc1: 100.0000 (77.3029) acc5: 100.0000 (93.4630) time: 0.0040 data: 0.0002 max mem: 86
Test: [25600/50000] eta: 0:01:42 loss: 0.3171 (0.9337) acc1: 100.0000 (77.3173) acc5: 100.0000 (93.4768) time: 0.0042 data: 0.0002 max mem: 86
Test: [25700/50000] eta: 0:01:41 loss: 0.9818 (0.9333) acc1: 0.0000 (77.2966) acc5: 100.0000 (93.4789) time: 0.0042 data: 0.0002 max mem: 86
Test: [25800/50000] eta: 0:01:41 loss: 0.7947 (0.9360) acc1: 0.0000 (77.2334) acc5: 100.0000 (93.4382) time: 0.0044 data: 0.0002 max mem: 86
Test: [25900/50000] eta: 0:01:41 loss: 0.1149 (0.9387) acc1: 100.0000 (77.1592) acc5: 100.0000 (93.4095) time: 0.0043 data: 0.0002 max mem: 86
Test: [26000/50000] eta: 0:01:40 loss: 0.0282 (0.9402) acc1: 100.0000 (77.1586) acc5: 100.0000 (93.3810) time: 0.0043 data: 0.0002 max mem: 86
Test: [26100/50000] eta: 0:01:40 loss: 0.0062 (0.9403) acc1: 100.0000 (77.1695) acc5: 100.0000 (93.3834) time: 0.0044 data: 0.0002 max mem: 86
Test: [26200/50000] eta: 0:01:39 loss: 0.8185 (0.9423) acc1: 100.0000 (77.1383) acc5: 100.0000 (93.3514) time: 0.0039 data: 0.0002 max mem: 86
Test: [26300/50000] eta: 0:01:39 loss: 0.0799 (0.9432) acc1: 100.0000 (77.0959) acc5: 100.0000 (93.3425) time: 0.0038 data: 0.0002 max mem: 86
Test: [26400/50000] eta: 0:01:39 loss: 1.7410 (0.9467) acc1: 0.0000 (76.9857) acc5: 100.0000 (93.3184) time: 0.0040 data: 0.0002 max mem: 86
Test: [26500/50000] eta: 0:01:38 loss: 0.3317 (0.9479) acc1: 100.0000 (76.9895) acc5: 100.0000 (93.2984) time: 0.0047 data: 0.0002 max mem: 86
Test: [26600/50000] eta: 0:01:38 loss: 0.2205 (0.9518) acc1: 100.0000 (76.9294) acc5: 100.0000 (93.2559) time: 0.0049 data: 0.0002 max mem: 86
Test: [26700/50000] eta: 0:01:37 loss: 0.0376 (0.9516) acc1: 100.0000 (76.9409) acc5: 100.0000 (93.2474) time: 0.0048 data: 0.0002 max mem: 86
Test: [26800/50000] eta: 0:01:37 loss: 0.0017 (0.9525) acc1: 100.0000 (76.9412) acc5: 100.0000 (93.2279) time: 0.0048 data: 0.0002 max mem: 86
Test: [26900/50000] eta: 0:01:37 loss: 0.0244 (0.9534) acc1: 100.0000 (76.9228) acc5: 100.0000 (93.2307) time: 0.0047 data: 0.0002 max mem: 86
Test: [27000/50000] eta: 0:01:36 loss: 0.3235 (0.9550) acc1: 100.0000 (76.8823) acc5: 100.0000 (93.2225) time: 0.0048 data: 0.0002 max mem: 86
Test: [27100/50000] eta: 0:01:36 loss: 1.2845 (0.9566) acc1: 0.0000 (76.8348) acc5: 100.0000 (93.1995) time: 0.0046 data: 0.0002 max mem: 86
Test: [27200/50000] eta: 0:01:36 loss: 0.4835 (0.9589) acc1: 100.0000 (76.7876) acc5: 100.0000 (93.1730) time: 0.0045 data: 0.0002 max mem: 86
Test: [27300/50000] eta: 0:01:35 loss: 0.0237 (0.9588) acc1: 100.0000 (76.7847) acc5: 100.0000 (93.1578) time: 0.0044 data: 0.0002 max mem: 86
Test: [27400/50000] eta: 0:01:35 loss: 0.0135 (0.9583) acc1: 100.0000 (76.8038) acc5: 100.0000 (93.1572) time: 0.0044 data: 0.0002 max mem: 86
Test: [27500/50000] eta: 0:01:34 loss: 1.2116 (0.9596) acc1: 0.0000 (76.7863) acc5: 100.0000 (93.1457) time: 0.0044 data: 0.0002 max mem: 86
Test: [27600/50000] eta: 0:01:34 loss: 0.1187 (0.9596) acc1: 100.0000 (76.7834) acc5: 100.0000 (93.1452) time: 0.0044 data: 0.0002 max mem: 86
Test: [27700/50000] eta: 0:01:34 loss: 0.0066 (0.9588) acc1: 100.0000 (76.7987) acc5: 100.0000 (93.1447) time: 0.0043 data: 0.0002 max mem: 86
Test: [27800/50000] eta: 0:01:33 loss: 0.0399 (0.9574) acc1: 100.0000 (76.8426) acc5: 100.0000 (93.1549) time: 0.0044 data: 0.0002 max mem: 86
Test: [27900/50000] eta: 0:01:33 loss: 0.1268 (0.9612) acc1: 100.0000 (76.8073) acc5: 100.0000 (93.1149) time: 0.0045 data: 0.0002 max mem: 86
Test: [28000/50000] eta: 0:01:32 loss: 0.0496 (0.9647) acc1: 100.0000 (76.7223) acc5: 100.0000 (93.0824) time: 0.0062 data: 0.0002 max mem: 86
Test: [28100/50000] eta: 0:01:32 loss: 0.0062 (0.9628) acc1: 100.0000 (76.7695) acc5: 100.0000 (93.0928) time: 0.0040 data: 0.0002 max mem: 86
Test: [28200/50000] eta: 0:01:31 loss: 0.3135 (0.9621) acc1: 100.0000 (76.7810) acc5: 100.0000 (93.0854) time: 0.0037 data: 0.0002 max mem: 86
Test: [28300/50000] eta: 0:01:31 loss: 0.0000 (0.9602) acc1: 100.0000 (76.8135) acc5: 100.0000 (93.1063) time: 0.0041 data: 0.0002 max mem: 86
Test: [28400/50000] eta: 0:01:31 loss: 2.3275 (0.9628) acc1: 0.0000 (76.7508) acc5: 100.0000 (93.0777) time: 0.0042 data: 0.0002 max mem: 86
Test: [28500/50000] eta: 0:01:30 loss: 0.0125 (0.9610) acc1: 100.0000 (76.7903) acc5: 100.0000 (93.0880) time: 0.0036 data: 0.0002 max mem: 86
Test: [28600/50000] eta: 0:01:30 loss: 0.0271 (0.9603) acc1: 100.0000 (76.7945) acc5: 100.0000 (93.0981) time: 0.0039 data: 0.0002 max mem: 86
Test: [28700/50000] eta: 0:01:29 loss: 0.0012 (0.9598) acc1: 100.0000 (76.8092) acc5: 100.0000 (93.1013) time: 0.0038 data: 0.0002 max mem: 86
Test: [28800/50000] eta: 0:01:29 loss: 0.0002 (0.9593) acc1: 100.0000 (76.8272) acc5: 100.0000 (93.1009) time: 0.0037 data: 0.0002 max mem: 86
Test: [28900/50000] eta: 0:01:28 loss: 0.5748 (0.9582) acc1: 100.0000 (76.8209) acc5: 100.0000 (93.1110) time: 0.0036 data: 0.0002 max mem: 86
Test: [29000/50000] eta: 0:01:28 loss: 0.0352 (0.9589) acc1: 100.0000 (76.7801) acc5: 100.0000 (93.1175) time: 0.0034 data: 0.0002 max mem: 86
Test: [29100/50000] eta: 0:01:27 loss: 0.0741 (0.9591) acc1: 100.0000 (76.7912) acc5: 100.0000 (93.1136) time: 0.0039 data: 0.0002 max mem: 86
Test: [29200/50000] eta: 0:01:27 loss: 0.0006 (0.9596) acc1: 100.0000 (76.7919) acc5: 100.0000 (93.1167) time: 0.0037 data: 0.0002 max mem: 86
Test: [29300/50000] eta: 0:01:26 loss: 2.7847 (0.9667) acc1: 0.0000 (76.6766) acc5: 100.0000 (93.0310) time: 0.0037 data: 0.0002 max mem: 86
Test: [29400/50000] eta: 0:01:26 loss: 0.4571 (0.9681) acc1: 100.0000 (76.6403) acc5: 100.0000 (93.0104) time: 0.0037 data: 0.0002 max mem: 86
Test: [29500/50000] eta: 0:01:26 loss: 0.0102 (0.9711) acc1: 100.0000 (76.5838) acc5: 100.0000 (92.9833) time: 0.0037 data: 0.0002 max mem: 86
Test: [29600/50000] eta: 0:01:25 loss: 0.0034 (0.9734) acc1: 100.0000 (76.5413) acc5: 100.0000 (92.9665) time: 0.0102 data: 0.0067 max mem: 86
Test: [29700/50000] eta: 0:01:25 loss: 0.1565 (0.9735) acc1: 100.0000 (76.5462) acc5: 100.0000 (92.9632) time: 0.0040 data: 0.0003 max mem: 86
Test: [29800/50000] eta: 0:01:24 loss: 0.0976 (0.9718) acc1: 100.0000 (76.5813) acc5: 100.0000 (92.9801) time: 0.0035 data: 0.0002 max mem: 86
Test: [29900/50000] eta: 0:01:24 loss: 0.0910 (0.9741) acc1: 100.0000 (76.5392) acc5: 100.0000 (92.9434) time: 0.0037 data: 0.0002 max mem: 86
Test: [30000/50000] eta: 0:01:23 loss: 0.0262 (0.9762) acc1: 100.0000 (76.5008) acc5: 100.0000 (92.9302) time: 0.0039 data: 0.0002 max mem: 86
Test: [30100/50000] eta: 0:01:23 loss: 0.4759 (0.9820) acc1: 100.0000 (76.4061) acc5: 100.0000 (92.8640) time: 0.0038 data: 0.0002 max mem: 86
Test: [30200/50000] eta: 0:01:23 loss: 0.0131 (0.9812) acc1: 100.0000 (76.4346) acc5: 100.0000 (92.8810) time: 0.0037 data: 0.0002 max mem: 86
Test: [30300/50000] eta: 0:01:22 loss: 0.0078 (0.9805) acc1: 100.0000 (76.4595) acc5: 100.0000 (92.8814) time: 0.0038 data: 0.0002 max mem: 86
Test: [30400/50000] eta: 0:01:22 loss: 0.0000 (0.9786) acc1: 100.0000 (76.5074) acc5: 100.0000 (92.8884) time: 0.0039 data: 0.0002 max mem: 86
Test: [30500/50000] eta: 0:01:21 loss: 0.0187 (0.9795) acc1: 100.0000 (76.4991) acc5: 100.0000 (92.8756) time: 0.0040 data: 0.0002 max mem: 86
Test: [30600/50000] eta: 0:01:21 loss: 0.0008 (0.9788) acc1: 100.0000 (76.5433) acc5: 100.0000 (92.8760) time: 0.0037 data: 0.0002 max mem: 86
Test: [30700/50000] eta: 0:01:20 loss: 0.0136 (0.9775) acc1: 100.0000 (76.5675) acc5: 100.0000 (92.8862) time: 0.0046 data: 0.0010 max mem: 86
Test: [30800/50000] eta: 0:01:20 loss: 0.0479 (0.9777) acc1: 100.0000 (76.5722) acc5: 100.0000 (92.8704) time: 0.0038 data: 0.0002 max mem: 86
Test: [30900/50000] eta: 0:01:19 loss: 0.2425 (0.9775) acc1: 100.0000 (76.5865) acc5: 100.0000 (92.8708) time: 0.0037 data: 0.0002 max mem: 86
Test: [31000/50000] eta: 0:01:19 loss: 0.6480 (0.9819) acc1: 100.0000 (76.4879) acc5: 100.0000 (92.8260) time: 0.0037 data: 0.0002 max mem: 86
Test: [31100/50000] eta: 0:01:19 loss: 0.0027 (0.9835) acc1: 100.0000 (76.4059) acc5: 100.0000 (92.8009) time: 0.0038 data: 0.0002 max mem: 86
Test: [31200/50000] eta: 0:01:18 loss: 2.3043 (0.9897) acc1: 0.0000 (76.3020) acc5: 100.0000 (92.7310) time: 0.0038 data: 0.0002 max mem: 86
Test: [31300/50000] eta: 0:01:18 loss: 0.0125 (0.9893) acc1: 100.0000 (76.3011) acc5: 100.0000 (92.7351) time: 0.0038 data: 0.0002 max mem: 86
Test: [31400/50000] eta: 0:01:17 loss: 0.0191 (0.9905) acc1: 100.0000 (76.2778) acc5: 100.0000 (92.7168) time: 0.0038 data: 0.0002 max mem: 86
Test: [31500/50000] eta: 0:01:17 loss: 0.1016 (0.9900) acc1: 100.0000 (76.2896) acc5: 100.0000 (92.7272) time: 0.0039 data: 0.0002 max mem: 86
Test: [31600/50000] eta: 0:01:16 loss: 0.9199 (0.9900) acc1: 100.0000 (76.2856) acc5: 100.0000 (92.7217) time: 0.0038 data: 0.0002 max mem: 86
Test: [31700/50000] eta: 0:01:16 loss: 5.9094 (0.9981) acc1: 0.0000 (76.1774) acc5: 0.0000 (92.6248) time: 0.0039 data: 0.0003 max mem: 86
Test: [31800/50000] eta: 0:01:16 loss: 0.5320 (0.9992) acc1: 100.0000 (76.1580) acc5: 100.0000 (92.6197) time: 0.0037 data: 0.0002 max mem: 86
Test: [31900/50000] eta: 0:01:15 loss: 0.0417 (0.9993) acc1: 100.0000 (76.1387) acc5: 100.0000 (92.6272) time: 0.0038 data: 0.0002 max mem: 86
Test: [32000/50000] eta: 0:01:15 loss: 1.1720 (1.0015) acc1: 0.0000 (75.9976) acc5: 100.0000 (92.6315) time: 0.0039 data: 0.0002 max mem: 86
Test: [32100/50000] eta: 0:01:14 loss: 0.0168 (1.0000) acc1: 100.0000 (76.0350) acc5: 100.0000 (92.6420) time: 0.0039 data: 0.0002 max mem: 86
Test: [32200/50000] eta: 0:01:14 loss: 1.3763 (1.0011) acc1: 0.0000 (75.9976) acc5: 100.0000 (92.6182) time: 0.0038 data: 0.0002 max mem: 86
Test: [32300/50000] eta: 0:01:13 loss: 0.0001 (1.0009) acc1: 100.0000 (76.0193) acc5: 100.0000 (92.6256) time: 0.0039 data: 0.0002 max mem: 86
Test: [32400/50000] eta: 0:01:13 loss: 0.4792 (1.0012) acc1: 100.0000 (76.0100) acc5: 100.0000 (92.6052) time: 0.0039 data: 0.0002 max mem: 86
Test: [32500/50000] eta: 0:01:12 loss: 0.1603 (1.0021) acc1: 100.0000 (76.0100) acc5: 100.0000 (92.5879) time: 0.0038 data: 0.0002 max mem: 86
Test: [32600/50000] eta: 0:01:12 loss: 0.8276 (1.0068) acc1: 0.0000 (75.9271) acc5: 100.0000 (92.5248) time: 0.0038 data: 0.0002 max mem: 86
Test: [32700/50000] eta: 0:01:12 loss: 0.0924 (1.0085) acc1: 100.0000 (75.9029) acc5: 100.0000 (92.4926) time: 0.0039 data: 0.0002 max mem: 86
Test: [32800/50000] eta: 0:01:11 loss: 1.1209 (1.0095) acc1: 100.0000 (75.8940) acc5: 100.0000 (92.4758) time: 0.0039 data: 0.0002 max mem: 86
Test: [32900/50000] eta: 0:01:11 loss: 0.7599 (1.0134) acc1: 0.0000 (75.7940) acc5: 100.0000 (92.4470) time: 0.0041 data: 0.0002 max mem: 86
Test: [33000/50000] eta: 0:01:10 loss: 0.0403 (1.0133) acc1: 100.0000 (75.8007) acc5: 100.0000 (92.4396) time: 0.0039 data: 0.0002 max mem: 86
Test: [33100/50000] eta: 0:01:10 loss: 0.0076 (1.0135) acc1: 100.0000 (75.8104) acc5: 100.0000 (92.4353) time: 0.0037 data: 0.0002 max mem: 86
Test: [33200/50000] eta: 0:01:09 loss: 0.9662 (1.0160) acc1: 0.0000 (75.7537) acc5: 100.0000 (92.4099) time: 0.0040 data: 0.0002 max mem: 86
Test: [33300/50000] eta: 0:01:09 loss: 0.3771 (1.0177) acc1: 100.0000 (75.6884) acc5: 100.0000 (92.4056) time: 0.0039 data: 0.0002 max mem: 86
Test: [33400/50000] eta: 0:01:09 loss: 1.1264 (1.0185) acc1: 0.0000 (75.6654) acc5: 100.0000 (92.3954) time: 0.0038 data: 0.0002 max mem: 86
Test: [33500/50000] eta: 0:01:08 loss: 0.0021 (1.0160) acc1: 100.0000 (75.7231) acc5: 100.0000 (92.4181) time: 0.0038 data: 0.0002 max mem: 86
Test: [33600/50000] eta: 0:01:08 loss: 0.0493 (1.0149) acc1: 100.0000 (75.7269) acc5: 100.0000 (92.4347) time: 0.0036 data: 0.0002 max mem: 86
Test: [33700/50000] eta: 0:01:07 loss: 1.4782 (1.0162) acc1: 0.0000 (75.7010) acc5: 100.0000 (92.4275) time: 0.0039 data: 0.0002 max mem: 86
Test: [33800/50000] eta: 0:01:07 loss: 2.0012 (1.0179) acc1: 0.0000 (75.6664) acc5: 100.0000 (92.4056) time: 0.0037 data: 0.0002 max mem: 86
Test: [33900/50000] eta: 0:01:06 loss: 0.2956 (1.0205) acc1: 100.0000 (75.6261) acc5: 100.0000 (92.3719) time: 0.0038 data: 0.0002 max mem: 86
Test: [34000/50000] eta: 0:01:06 loss: 0.0472 (1.0215) acc1: 100.0000 (75.6272) acc5: 100.0000 (92.3502) time: 0.0039 data: 0.0002 max mem: 86
Test: [34100/50000] eta: 0:01:06 loss: 1.8225 (1.0236) acc1: 0.0000 (75.5755) acc5: 100.0000 (92.3316) time: 0.0039 data: 0.0002 max mem: 86
Test: [34200/50000] eta: 0:01:05 loss: 0.0572 (1.0243) acc1: 100.0000 (75.5709) acc5: 100.0000 (92.3277) time: 0.0037 data: 0.0002 max mem: 86
Test: [34300/50000] eta: 0:01:05 loss: 0.0000 (1.0229) acc1: 100.0000 (75.6100) acc5: 100.0000 (92.3268) time: 0.0040 data: 0.0002 max mem: 86
Test: [34400/50000] eta: 0:01:04 loss: 0.0172 (1.0238) acc1: 100.0000 (75.5908) acc5: 100.0000 (92.3142) time: 0.0049 data: 0.0002 max mem: 86
Test: [34500/50000] eta: 0:01:04 loss: 2.5449 (1.0254) acc1: 0.0000 (75.5572) acc5: 100.0000 (92.2988) time: 0.0043 data: 0.0002 max mem: 86
Test: [34600/50000] eta: 0:01:03 loss: 0.1552 (1.0262) acc1: 100.0000 (75.5556) acc5: 100.0000 (92.2835) time: 0.0043 data: 0.0002 max mem: 86
Test: [34700/50000] eta: 0:01:03 loss: 0.4390 (1.0281) acc1: 100.0000 (75.4935) acc5: 100.0000 (92.2596) time: 0.0044 data: 0.0002 max mem: 86
Test: [34800/50000] eta: 0:01:03 loss: 0.0076 (1.0276) acc1: 100.0000 (75.5151) acc5: 100.0000 (92.2588) time: 0.0041 data: 0.0002 max mem: 86
Test: [34900/50000] eta: 0:01:02 loss: 0.7986 (1.0291) acc1: 100.0000 (75.4792) acc5: 100.0000 (92.2495) time: 0.0042 data: 0.0002 max mem: 86
Test: [35000/50000] eta: 0:01:02 loss: 0.0039 (1.0286) acc1: 100.0000 (75.4921) acc5: 100.0000 (92.2545) time: 0.0042 data: 0.0002 max mem: 86
Test: [35100/50000] eta: 0:01:01 loss: 0.0001 (1.0286) acc1: 100.0000 (75.5021) acc5: 100.0000 (92.2538) time: 0.0041 data: 0.0002 max mem: 86
Test: [35200/50000] eta: 0:01:01 loss: 0.0196 (1.0284) acc1: 100.0000 (75.4922) acc5: 100.0000 (92.2587) time: 0.0041 data: 0.0002 max mem: 86
Test: [35300/50000] eta: 0:01:01 loss: 0.8688 (1.0297) acc1: 0.0000 (75.4766) acc5: 100.0000 (92.2608) time: 0.0041 data: 0.0002 max mem: 86
Test: [35400/50000] eta: 0:01:00 loss: 0.0061 (1.0310) acc1: 100.0000 (75.4668) acc5: 100.0000 (92.2432) time: 0.0039 data: 0.0002 max mem: 86
Test: [35500/50000] eta: 0:01:00 loss: 0.3687 (1.0314) acc1: 100.0000 (75.4373) acc5: 100.0000 (92.2509) time: 0.0040 data: 0.0002 max mem: 86
Test: [35600/50000] eta: 0:00:59 loss: 0.0745 (1.0329) acc1: 100.0000 (75.3996) acc5: 100.0000 (92.2334) time: 0.0042 data: 0.0002 max mem: 86
Test: [35700/50000] eta: 0:00:59 loss: 0.0009 (1.0327) acc1: 100.0000 (75.4153) acc5: 100.0000 (92.2271) time: 0.0042 data: 0.0002 max mem: 86
Test: [35800/50000] eta: 0:00:58 loss: 0.0018 (1.0322) acc1: 100.0000 (75.4309) acc5: 100.0000 (92.2293) time: 0.0042 data: 0.0002 max mem: 86
Test: [35900/50000] eta: 0:00:58 loss: 0.3446 (1.0322) acc1: 100.0000 (75.4352) acc5: 100.0000 (92.2258) time: 0.0041 data: 0.0002 max mem: 86
Test: [36000/50000] eta: 0:00:58 loss: 0.0002 (1.0324) acc1: 100.0000 (75.4312) acc5: 100.0000 (92.2252) time: 0.0040 data: 0.0002 max mem: 86
Test: [36100/50000] eta: 0:00:57 loss: 0.0874 (1.0323) acc1: 100.0000 (75.4411) acc5: 100.0000 (92.2246) time: 0.0044 data: 0.0002 max mem: 86
Test: [36200/50000] eta: 0:00:57 loss: 0.0008 (1.0320) acc1: 100.0000 (75.4620) acc5: 100.0000 (92.2157) time: 0.0041 data: 0.0002 max mem: 86
Test: [36300/50000] eta: 0:00:56 loss: 1.0876 (1.0326) acc1: 0.0000 (75.4332) acc5: 100.0000 (92.2068) time: 0.0045 data: 0.0002 max mem: 86
Test: [36400/50000] eta: 0:00:56 loss: 0.0012 (1.0310) acc1: 100.0000 (75.4677) acc5: 100.0000 (92.2172) time: 0.0044 data: 0.0002 max mem: 86
Test: [36500/50000] eta: 0:00:56 loss: 0.1963 (1.0367) acc1: 100.0000 (75.3925) acc5: 100.0000 (92.1372) time: 0.0045 data: 0.0002 max mem: 86
Test: [36600/50000] eta: 0:00:55 loss: 0.8977 (1.0400) acc1: 0.0000 (75.3367) acc5: 100.0000 (92.1013) time: 0.0043 data: 0.0002 max mem: 86
Test: [36700/50000] eta: 0:00:55 loss: 1.2821 (1.0421) acc1: 100.0000 (75.3031) acc5: 100.0000 (92.0765) time: 0.0044 data: 0.0002 max mem: 86
Test: [36800/50000] eta: 0:00:54 loss: 0.3187 (1.0422) acc1: 100.0000 (75.2860) acc5: 100.0000 (92.0790) time: 0.0050 data: 0.0002 max mem: 86
Test: [36900/50000] eta: 0:00:54 loss: 0.5414 (1.0421) acc1: 100.0000 (75.2852) acc5: 100.0000 (92.0761) time: 0.0044 data: 0.0002 max mem: 86
Test: [37000/50000] eta: 0:00:54 loss: 0.0000 (1.0410) acc1: 100.0000 (75.3142) acc5: 100.0000 (92.0759) time: 0.0043 data: 0.0002 max mem: 86
Test: [37100/50000] eta: 0:00:53 loss: 0.4119 (1.0430) acc1: 100.0000 (75.2594) acc5: 100.0000 (92.0514) time: 0.0044 data: 0.0002 max mem: 86
Test: [37200/50000] eta: 0:00:53 loss: 0.5268 (1.0443) acc1: 100.0000 (75.2292) acc5: 100.0000 (92.0352) time: 0.0044 data: 0.0002 max mem: 86
Test: [37300/50000] eta: 0:00:52 loss: 0.0056 (1.0465) acc1: 100.0000 (75.1588) acc5: 100.0000 (92.0243) time: 0.0045 data: 0.0002 max mem: 86
Test: [37400/50000] eta: 0:00:52 loss: 1.4897 (1.0469) acc1: 0.0000 (75.1611) acc5: 100.0000 (92.0029) time: 0.0043 data: 0.0002 max mem: 86
Test: [37500/50000] eta: 0:00:52 loss: 0.0313 (1.0487) acc1: 100.0000 (75.1100) acc5: 100.0000 (91.9815) time: 0.0043 data: 0.0002 max mem: 86
Test: [37600/50000] eta: 0:00:51 loss: 0.3074 (1.0492) acc1: 100.0000 (75.0911) acc5: 100.0000 (91.9869) time: 0.0044 data: 0.0002 max mem: 86
Test: [37700/50000] eta: 0:00:51 loss: 0.0172 (1.0496) acc1: 100.0000 (75.0882) acc5: 100.0000 (91.9790) time: 0.0044 data: 0.0002 max mem: 86
Test: [37800/50000] eta: 0:00:50 loss: 0.0022 (1.0492) acc1: 100.0000 (75.0985) acc5: 100.0000 (91.9817) time: 0.0044 data: 0.0002 max mem: 86
Test: [37900/50000] eta: 0:00:50 loss: 0.0170 (1.0489) acc1: 100.0000 (75.1036) acc5: 100.0000 (91.9765) time: 0.0044 data: 0.0002 max mem: 86
Test: [38000/50000] eta: 0:00:49 loss: 0.0966 (1.0499) acc1: 100.0000 (75.0954) acc5: 100.0000 (91.9502) time: 0.0044 data: 0.0002 max mem: 86
Test: [38100/50000] eta: 0:00:49 loss: 0.0094 (1.0497) acc1: 100.0000 (75.0978) acc5: 100.0000 (91.9556) time: 0.0044 data: 0.0002 max mem: 86
Test: [38200/50000] eta: 0:00:49 loss: 0.0218 (1.0517) acc1: 100.0000 (75.0792) acc5: 100.0000 (91.9191) time: 0.0044 data: 0.0002 max mem: 86
Test: [38300/50000] eta: 0:00:48 loss: 0.0913 (1.0534) acc1: 100.0000 (75.0503) acc5: 100.0000 (91.9010) time: 0.0040 data: 0.0002 max mem: 86
Test: [38400/50000] eta: 0:00:48 loss: 3.0212 (1.0541) acc1: 0.0000 (75.0475) acc5: 100.0000 (91.8830) time: 0.0040 data: 0.0002 max mem: 86
Test: [38500/50000] eta: 0:00:47 loss: 0.0590 (1.0545) acc1: 100.0000 (75.0500) acc5: 100.0000 (91.8677) time: 0.0040 data: 0.0002 max mem: 86
Test: [38600/50000] eta: 0:00:47 loss: 0.0100 (1.0544) acc1: 100.0000 (75.0706) acc5: 100.0000 (91.8681) time: 0.0043 data: 0.0001 max mem: 86
Test: [38700/50000] eta: 0:00:47 loss: 0.4904 (1.0564) acc1: 100.0000 (75.0239) acc5: 100.0000 (91.8478) time: 0.0044 data: 0.0002 max mem: 86
Test: [38800/50000] eta: 0:00:46 loss: 0.4960 (1.0580) acc1: 100.0000 (74.9852) acc5: 100.0000 (91.8250) time: 0.0045 data: 0.0002 max mem: 86
Test: [38900/50000] eta: 0:00:46 loss: 0.0769 (1.0580) acc1: 100.0000 (74.9852) acc5: 100.0000 (91.8305) time: 0.0044 data: 0.0002 max mem: 86
Test: [39000/50000] eta: 0:00:45 loss: 0.0051 (1.0587) acc1: 100.0000 (74.9904) acc5: 100.0000 (91.8207) time: 0.0045 data: 0.0002 max mem: 86
Test: [39100/50000] eta: 0:00:45 loss: 0.0018 (1.0566) acc1: 100.0000 (75.0313) acc5: 100.0000 (91.8391) time: 0.0043 data: 0.0002 max mem: 86
Test: [39200/50000] eta: 0:00:45 loss: 0.0012 (1.0582) acc1: 100.0000 (74.9802) acc5: 100.0000 (91.8242) time: 0.0042 data: 0.0002 max mem: 86
Test: [39300/50000] eta: 0:00:44 loss: 0.1853 (1.0605) acc1: 100.0000 (74.9396) acc5: 100.0000 (91.7865) time: 0.0041 data: 0.0002 max mem: 86
Test: [39400/50000] eta: 0:00:44 loss: 0.6406 (1.0621) acc1: 0.0000 (74.9042) acc5: 100.0000 (91.7667) time: 0.0042 data: 0.0002 max mem: 86
Test: [39500/50000] eta: 0:00:43 loss: 0.1936 (1.0617) acc1: 100.0000 (74.9095) acc5: 100.0000 (91.7698) time: 0.0042 data: 0.0002 max mem: 86
Test: [39600/50000] eta: 0:00:43 loss: 0.0425 (1.0628) acc1: 100.0000 (74.8946) acc5: 100.0000 (91.7527) time: 0.0041 data: 0.0002 max mem: 86
Test: [39700/50000] eta: 0:00:42 loss: 0.0153 (1.0645) acc1: 100.0000 (74.8747) acc5: 100.0000 (91.7357) time: 0.0042 data: 0.0002 max mem: 86
Test: [39800/50000] eta: 0:00:42 loss: 0.0769 (1.0651) acc1: 100.0000 (74.8675) acc5: 100.0000 (91.7238) time: 0.0041 data: 0.0002 max mem: 86
Test: [39900/50000] eta: 0:00:42 loss: 0.0277 (1.0658) acc1: 100.0000 (74.8528) acc5: 100.0000 (91.7170) time: 0.0051 data: 0.0002 max mem: 86
Test: [40000/50000] eta: 0:00:41 loss: 0.6600 (1.0687) acc1: 100.0000 (74.8131) acc5: 100.0000 (91.6777) time: 0.0047 data: 0.0002 max mem: 86
Test: [40100/50000] eta: 0:00:41 loss: 0.0090 (1.0668) acc1: 100.0000 (74.8560) acc5: 100.0000 (91.6935) time: 0.0043 data: 0.0002 max mem: 86
Test: [40200/50000] eta: 0:00:40 loss: 0.0264 (1.0652) acc1: 100.0000 (74.8887) acc5: 100.0000 (91.7067) time: 0.0043 data: 0.0002 max mem: 86
Test: [40300/50000] eta: 0:00:40 loss: 0.0093 (1.0654) acc1: 100.0000 (74.8815) acc5: 100.0000 (91.7099) time: 0.0043 data: 0.0002 max mem: 86
Test: [40400/50000] eta: 0:00:40 loss: 0.0218 (1.0666) acc1: 100.0000 (74.8744) acc5: 100.0000 (91.6834) time: 0.0085 data: 0.0045 max mem: 86
Test: [40500/50000] eta: 0:00:39 loss: 1.6256 (1.0685) acc1: 0.0000 (74.8228) acc5: 100.0000 (91.6644) time: 0.0046 data: 0.0002 max mem: 86
Test: [40600/50000] eta: 0:00:39 loss: 0.1739 (1.0713) acc1: 100.0000 (74.7691) acc5: 100.0000 (91.6381) time: 0.0045 data: 0.0002 max mem: 86
Test: [40700/50000] eta: 0:00:38 loss: 2.0720 (1.0730) acc1: 0.0000 (74.7353) acc5: 100.0000 (91.6145) time: 0.0045 data: 0.0002 max mem: 86
Test: [40800/50000] eta: 0:00:38 loss: 0.2095 (1.0723) acc1: 100.0000 (74.7555) acc5: 100.0000 (91.6277) time: 0.0041 data: 0.0002 max mem: 86
Test: [40900/50000] eta: 0:00:38 loss: 0.9667 (1.0716) acc1: 100.0000 (74.7610) acc5: 100.0000 (91.6359) time: 0.0040 data: 0.0002 max mem: 86
Test: [41000/50000] eta: 0:00:37 loss: 0.1690 (1.0745) acc1: 100.0000 (74.7128) acc5: 100.0000 (91.6051) time: 0.0040 data: 0.0002 max mem: 86
Test: [41100/50000] eta: 0:00:37 loss: 0.0553 (1.0729) acc1: 100.0000 (74.7500) acc5: 100.0000 (91.6182) time: 0.0041 data: 0.0002 max mem: 86
Test: [41200/50000] eta: 0:00:36 loss: 0.7887 (1.0736) acc1: 100.0000 (74.7506) acc5: 100.0000 (91.5973) time: 0.0042 data: 0.0002 max mem: 86
Test: [41300/50000] eta: 0:00:36 loss: 0.0577 (1.0742) acc1: 100.0000 (74.7125) acc5: 100.0000 (91.5958) time: 0.0040 data: 0.0002 max mem: 86
Test: [41400/50000] eta: 0:00:35 loss: 0.4766 (1.0766) acc1: 100.0000 (74.6649) acc5: 100.0000 (91.5751) time: 0.0041 data: 0.0002 max mem: 86
Test: [41500/50000] eta: 0:00:35 loss: 0.0224 (1.0772) acc1: 100.0000 (74.6633) acc5: 100.0000 (91.5616) time: 0.0041 data: 0.0002 max mem: 86
Test: [41600/50000] eta: 0:00:35 loss: 0.2120 (1.0775) acc1: 100.0000 (74.6569) acc5: 100.0000 (91.5579) time: 0.0041 data: 0.0002 max mem: 86
Test: [41700/50000] eta: 0:00:34 loss: 0.0137 (1.0766) acc1: 100.0000 (74.6745) acc5: 100.0000 (91.5661) time: 0.0040 data: 0.0001 max mem: 86
Test: [41800/50000] eta: 0:00:34 loss: 0.0237 (1.0781) acc1: 0.0000 (74.6322) acc5: 100.0000 (91.5409) time: 0.0041 data: 0.0002 max mem: 86
Test: [41900/50000] eta: 0:00:33 loss: 2.3601 (1.0831) acc1: 0.0000 (74.5113) acc5: 100.0000 (91.4775) time: 0.0041 data: 0.0002 max mem: 86
Test: [42000/50000] eta: 0:00:33 loss: 0.3241 (1.0853) acc1: 100.0000 (74.4601) acc5: 100.0000 (91.4597) time: 0.0041 data: 0.0002 max mem: 86
Test: [42100/50000] eta: 0:00:33 loss: 0.9063 (1.0870) acc1: 100.0000 (74.4329) acc5: 100.0000 (91.4349) time: 0.0041 data: 0.0002 max mem: 86
Test: [42200/50000] eta: 0:00:32 loss: 0.0135 (1.0873) acc1: 100.0000 (74.4129) acc5: 100.0000 (91.4291) time: 0.0040 data: 0.0002 max mem: 86
Test: [42300/50000] eta: 0:00:32 loss: 0.1794 (1.0892) acc1: 100.0000 (74.3883) acc5: 100.0000 (91.4045) time: 0.0040 data: 0.0002 max mem: 86
Test: [42400/50000] eta: 0:00:31 loss: 0.0941 (1.0898) acc1: 100.0000 (74.3615) acc5: 100.0000 (91.3988) time: 0.0040 data: 0.0002 max mem: 86
Test: [42500/50000] eta: 0:00:31 loss: 0.0792 (1.0902) acc1: 100.0000 (74.3347) acc5: 100.0000 (91.4073) time: 0.0040 data: 0.0002 max mem: 86
Test: [42600/50000] eta: 0:00:30 loss: 0.0957 (1.0900) acc1: 100.0000 (74.3292) acc5: 100.0000 (91.4133) time: 0.0056 data: 0.0002 max mem: 86
Test: [42700/50000] eta: 0:00:30 loss: 0.0044 (1.0891) acc1: 100.0000 (74.3472) acc5: 100.0000 (91.4217) time: 0.0041 data: 0.0002 max mem: 86
Test: [42800/50000] eta: 0:00:30 loss: 0.0020 (1.0896) acc1: 100.0000 (74.3487) acc5: 100.0000 (91.4091) time: 0.0041 data: 0.0002 max mem: 86
Test: [42900/50000] eta: 0:00:29 loss: 0.2166 (1.0903) acc1: 100.0000 (74.3293) acc5: 100.0000 (91.4105) time: 0.0041 data: 0.0002 max mem: 86
Test: [43000/50000] eta: 0:00:29 loss: 0.4386 (1.0907) acc1: 100.0000 (74.3239) acc5: 100.0000 (91.3909) time: 0.0043 data: 0.0002 max mem: 86
Test: [43100/50000] eta: 0:00:28 loss: 0.1530 (1.0923) acc1: 100.0000 (74.2883) acc5: 100.0000 (91.3668) time: 0.0040 data: 0.0002 max mem: 86
Test: [43200/50000] eta: 0:00:28 loss: 0.0065 (1.0916) acc1: 100.0000 (74.3085) acc5: 100.0000 (91.3706) time: 0.0041 data: 0.0002 max mem: 86
Test: [43300/50000] eta: 0:00:27 loss: 0.5499 (1.0922) acc1: 100.0000 (74.2685) acc5: 100.0000 (91.3651) time: 0.0040 data: 0.0002 max mem: 86
Test: [43400/50000] eta: 0:00:27 loss: 0.3370 (1.0916) acc1: 100.0000 (74.2748) acc5: 100.0000 (91.3735) time: 0.0041 data: 0.0002 max mem: 86
Test: [43500/50000] eta: 0:00:27 loss: 0.2863 (1.0939) acc1: 100.0000 (74.2374) acc5: 100.0000 (91.3519) time: 0.0041 data: 0.0002 max mem: 86
Test: [43600/50000] eta: 0:00:26 loss: 0.2812 (1.0930) acc1: 100.0000 (74.2391) acc5: 100.0000 (91.3672) time: 0.0041 data: 0.0002 max mem: 86
Test: [43700/50000] eta: 0:00:26 loss: 0.0013 (1.0925) acc1: 100.0000 (74.2592) acc5: 100.0000 (91.3663) time: 0.0041 data: 0.0002 max mem: 86
Test: [43800/50000] eta: 0:00:25 loss: 0.1880 (1.0916) acc1: 100.0000 (74.2791) acc5: 100.0000 (91.3723) time: 0.0041 data: 0.0002 max mem: 86
Test: [43900/50000] eta: 0:00:25 loss: 0.0064 (1.0920) acc1: 100.0000 (74.2375) acc5: 100.0000 (91.3692) time: 0.0040 data: 0.0002 max mem: 86
Test: [44000/50000] eta: 0:00:25 loss: 0.0775 (1.0924) acc1: 100.0000 (74.2165) acc5: 100.0000 (91.3638) time: 0.0040 data: 0.0002 max mem: 86
Test: [44100/50000] eta: 0:00:24 loss: 0.1228 (1.0920) acc1: 100.0000 (74.2341) acc5: 100.0000 (91.3675) time: 0.0042 data: 0.0002 max mem: 86
Test: [44200/50000] eta: 0:00:24 loss: 1.8513 (1.0932) acc1: 0.0000 (74.2110) acc5: 100.0000 (91.3509) time: 0.0045 data: 0.0002 max mem: 86
Test: [44300/50000] eta: 0:00:23 loss: 3.8736 (1.0965) acc1: 0.0000 (74.1541) acc5: 100.0000 (91.3117) time: 0.0044 data: 0.0002 max mem: 86
Test: [44400/50000] eta: 0:00:23 loss: 0.4045 (1.0968) acc1: 100.0000 (74.1650) acc5: 100.0000 (91.3065) time: 0.0044 data: 0.0002 max mem: 86
Test: [44500/50000] eta: 0:00:22 loss: 0.4718 (1.0967) acc1: 100.0000 (74.1579) acc5: 100.0000 (91.3036) time: 0.0051 data: 0.0002 max mem: 86
Test: [44600/50000] eta: 0:00:22 loss: 0.0673 (1.0961) acc1: 100.0000 (74.1732) acc5: 100.0000 (91.3051) time: 0.0049 data: 0.0002 max mem: 86
Test: [44700/50000] eta: 0:00:22 loss: 1.4266 (1.0967) acc1: 0.0000 (74.1370) acc5: 100.0000 (91.3089) time: 0.0046 data: 0.0002 max mem: 86
Test: [44800/50000] eta: 0:00:21 loss: 0.0431 (1.0960) acc1: 100.0000 (74.1479) acc5: 100.0000 (91.3216) time: 0.0043 data: 0.0002 max mem: 86
Test: [44900/50000] eta: 0:00:21 loss: 0.0027 (1.0956) acc1: 100.0000 (74.1632) acc5: 100.0000 (91.3276) time: 0.0042 data: 0.0002 max mem: 86
Test: [45000/50000] eta: 0:00:20 loss: 2.0405 (1.0988) acc1: 0.0000 (74.0917) acc5: 100.0000 (91.3024) time: 0.0041 data: 0.0002 max mem: 86
Test: [45100/50000] eta: 0:00:20 loss: 0.0499 (1.0986) acc1: 100.0000 (74.1026) acc5: 100.0000 (91.2907) time: 0.0046 data: 0.0002 max mem: 86
Test: [45200/50000] eta: 0:00:20 loss: 0.0301 (1.0984) acc1: 100.0000 (74.1112) acc5: 100.0000 (91.2944) time: 0.0043 data: 0.0002 max mem: 86
Test: [45300/50000] eta: 0:00:19 loss: 0.2390 (1.0996) acc1: 100.0000 (74.0889) acc5: 100.0000 (91.2629) time: 0.0042 data: 0.0002 max mem: 86
Test: [45400/50000] eta: 0:00:19 loss: 1.2460 (1.1034) acc1: 0.0000 (74.0006) acc5: 100.0000 (91.2227) time: 0.0042 data: 0.0002 max mem: 86
Test: [45500/50000] eta: 0:00:18 loss: 0.6810 (1.1048) acc1: 100.0000 (73.9588) acc5: 100.0000 (91.2266) time: 0.0043 data: 0.0002 max mem: 86
Test: [45600/50000] eta: 0:00:18 loss: 0.2011 (1.1071) acc1: 100.0000 (73.9194) acc5: 100.0000 (91.2041) time: 0.0043 data: 0.0002 max mem: 86
Test: [45700/50000] eta: 0:00:17 loss: 0.0228 (1.1068) acc1: 100.0000 (73.9240) acc5: 100.0000 (91.2059) time: 0.0043 data: 0.0002 max mem: 86
Test: [45800/50000] eta: 0:00:17 loss: 0.0003 (1.1058) acc1: 100.0000 (73.9416) acc5: 100.0000 (91.2185) time: 0.0041 data: 0.0002 max mem: 86
Test: [45900/50000] eta: 0:00:17 loss: 0.1972 (1.1048) acc1: 100.0000 (73.9657) acc5: 100.0000 (91.2289) time: 0.0042 data: 0.0002 max mem: 86
Test: [46000/50000] eta: 0:00:16 loss: 0.1110 (1.1049) acc1: 100.0000 (73.9745) acc5: 100.0000 (91.2176) time: 0.0043 data: 0.0002 max mem: 86
Test: [46100/50000] eta: 0:00:16 loss: 0.6403 (1.1047) acc1: 100.0000 (73.9745) acc5: 100.0000 (91.2236) time: 0.0043 data: 0.0002 max mem: 86
Test: [46200/50000] eta: 0:00:15 loss: 1.0495 (1.1057) acc1: 100.0000 (73.9703) acc5: 100.0000 (91.2145) time: 0.0043 data: 0.0002 max mem: 86
Test: [46300/50000] eta: 0:00:15 loss: 0.4804 (1.1053) acc1: 100.0000 (73.9595) acc5: 100.0000 (91.2183) time: 0.0042 data: 0.0002 max mem: 86
Test: [46400/50000] eta: 0:00:15 loss: 0.0027 (1.1038) acc1: 100.0000 (73.9812) acc5: 100.0000 (91.2351) time: 0.0045 data: 0.0002 max mem: 86
Test: [46500/50000] eta: 0:00:14 loss: 0.0935 (1.1043) acc1: 100.0000 (73.9575) acc5: 100.0000 (91.2346) time: 0.0040 data: 0.0002 max mem: 86
Test: [46600/50000] eta: 0:00:14 loss: 0.0400 (1.1054) acc1: 100.0000 (73.9297) acc5: 100.0000 (91.2255) time: 0.0039 data: 0.0002 max mem: 86
Test: [46700/50000] eta: 0:00:13 loss: 0.0019 (1.1049) acc1: 100.0000 (73.9406) acc5: 100.0000 (91.2207) time: 0.0040 data: 0.0002 max mem: 86
Test: [46800/50000] eta: 0:00:13 loss: 0.0435 (1.1051) acc1: 100.0000 (73.9429) acc5: 100.0000 (91.2160) time: 0.0040 data: 0.0002 max mem: 86
Test: [46900/50000] eta: 0:00:12 loss: 0.0004 (1.1039) acc1: 100.0000 (73.9707) acc5: 100.0000 (91.2283) time: 0.0064 data: 0.0026 max mem: 86
Test: [47000/50000] eta: 0:00:12 loss: 0.0368 (1.1030) acc1: 100.0000 (73.9771) acc5: 100.0000 (91.2470) time: 0.0040 data: 0.0002 max mem: 86
Test: [47100/50000] eta: 0:00:12 loss: 0.0096 (1.1024) acc1: 100.0000 (73.9814) acc5: 100.0000 (91.2550) time: 0.0041 data: 0.0002 max mem: 86
Test: [47200/50000] eta: 0:00:11 loss: 0.1077 (1.1020) acc1: 100.0000 (73.9772) acc5: 100.0000 (91.2608) time: 0.0041 data: 0.0002 max mem: 86
Test: [47300/50000] eta: 0:00:11 loss: 0.0041 (1.1010) acc1: 100.0000 (73.9984) acc5: 100.0000 (91.2729) time: 0.0041 data: 0.0002 max mem: 86
Test: [47400/50000] eta: 0:00:10 loss: 0.6002 (1.1012) acc1: 100.0000 (73.9816) acc5: 100.0000 (91.2871) time: 0.0040 data: 0.0002 max mem: 86
Test: [47500/50000] eta: 0:00:10 loss: 0.0593 (1.1010) acc1: 100.0000 (73.9816) acc5: 100.0000 (91.2907) time: 0.0041 data: 0.0002 max mem: 86
Test: [47600/50000] eta: 0:00:10 loss: 0.1147 (1.1010) acc1: 100.0000 (73.9795) acc5: 100.0000 (91.2985) time: 0.0041 data: 0.0002 max mem: 86
Test: [47700/50000] eta: 0:00:09 loss: 0.0013 (1.0998) acc1: 100.0000 (74.0068) acc5: 100.0000 (91.3042) time: 0.0042 data: 0.0002 max mem: 86
Test: [47800/50000] eta: 0:00:09 loss: 0.0000 (1.0988) acc1: 100.0000 (74.0298) acc5: 100.0000 (91.3140) time: 0.0041 data: 0.0002 max mem: 86
Test: [47900/50000] eta: 0:00:08 loss: 0.0004 (1.0977) acc1: 100.0000 (74.0590) acc5: 100.0000 (91.3259) time: 0.0041 data: 0.0002 max mem: 86
Test: [48000/50000] eta: 0:00:08 loss: 0.0048 (1.0966) acc1: 100.0000 (74.0818) acc5: 100.0000 (91.3335) time: 0.0041 data: 0.0002 max mem: 86
Test: [48100/50000] eta: 0:00:07 loss: 1.0288 (1.0997) acc1: 100.0000 (74.0276) acc5: 100.0000 (91.2996) time: 0.0041 data: 0.0002 max mem: 86
Test: [48200/50000] eta: 0:00:07 loss: 0.0181 (1.0999) acc1: 100.0000 (74.0296) acc5: 100.0000 (91.2927) time: 0.0041 data: 0.0002 max mem: 86
Test: [48300/50000] eta: 0:00:07 loss: 0.0095 (1.0998) acc1: 100.0000 (74.0399) acc5: 100.0000 (91.2921) time: 0.0044 data: 0.0002 max mem: 86
Test: [48400/50000] eta: 0:00:06 loss: 0.1439 (1.1010) acc1: 100.0000 (74.0171) acc5: 100.0000 (91.2874) time: 0.0041 data: 0.0002 max mem: 86
Test: [48500/50000] eta: 0:00:06 loss: 2.2339 (1.1049) acc1: 0.0000 (73.9449) acc5: 100.0000 (91.2435) time: 0.0043 data: 0.0002 max mem: 86
Test: [48600/50000] eta: 0:00:05 loss: 0.0188 (1.1055) acc1: 100.0000 (73.9244) acc5: 100.0000 (91.2409) time: 0.0041 data: 0.0002 max mem: 86
Test: [48700/50000] eta: 0:00:05 loss: 0.1100 (1.1054) acc1: 100.0000 (73.9163) acc5: 100.0000 (91.2527) time: 0.0041 data: 0.0002 max mem: 86
Test: [48800/50000] eta: 0:00:05 loss: 1.7498 (1.1053) acc1: 0.0000 (73.9206) acc5: 100.0000 (91.2584) time: 0.0042 data: 0.0002 max mem: 86
Test: [48900/50000] eta: 0:00:04 loss: 0.3690 (1.1059) acc1: 100.0000 (73.8983) acc5: 100.0000 (91.2660) time: 0.0043 data: 0.0002 max mem: 86
Test: [49000/50000] eta: 0:00:04 loss: 0.1133 (1.1072) acc1: 100.0000 (73.8699) acc5: 100.0000 (91.2573) time: 0.0042 data: 0.0002 max mem: 86
Test: [49100/50000] eta: 0:00:03 loss: 0.1942 (1.1067) acc1: 100.0000 (73.8906) acc5: 100.0000 (91.2609) time: 0.0046 data: 0.0003 max mem: 86
Test: [49200/50000] eta: 0:00:03 loss: 0.0112 (1.1064) acc1: 100.0000 (73.9030) acc5: 100.0000 (91.2624) time: 0.0044 data: 0.0002 max mem: 86
Test: [49300/50000] eta: 0:00:02 loss: 0.0015 (1.1046) acc1: 100.0000 (73.9478) acc5: 100.0000 (91.2760) time: 0.0042 data: 0.0001 max mem: 86
Test: [49400/50000] eta: 0:00:02 loss: 0.6341 (1.1039) acc1: 0.0000 (73.9459) acc5: 100.0000 (91.2897) time: 0.0044 data: 0.0002 max mem: 86
Test: [49500/50000] eta: 0:00:02 loss: 0.0015 (1.1020) acc1: 100.0000 (73.9884) acc5: 100.0000 (91.3052) time: 0.0043 data: 0.0002 max mem: 86
Test: [49600/50000] eta: 0:00:01 loss: 0.0005 (1.1002) acc1: 100.0000 (74.0287) acc5: 100.0000 (91.3207) time: 0.0044 data: 0.0002 max mem: 86
Test: [49700/50000] eta: 0:00:01 loss: 0.0000 (1.0989) acc1: 100.0000 (74.0629) acc5: 100.0000 (91.3342) time: 0.0043 data: 0.0002 max mem: 86
Test: [49800/50000] eta: 0:00:00 loss: 0.0000 (1.0973) acc1: 100.0000 (74.1049) acc5: 100.0000 (91.3496) time: 0.0042 data: 0.0002 max mem: 86
Test: [49900/50000] eta: 0:00:00 loss: 0.5711 (1.0966) acc1: 100.0000 (74.1188) acc5: 100.0000 (91.3589) time: 0.0043 data: 0.0002 max mem: 86
Test: Total time: 0:03:30
* Acc@1 74.054 Acc@5 91.340
utils.py 0 → 100644
View file @ cea8529d
from collections import defaultdict, deque, OrderedDict
import copy
import datetime
import hashlib
import time
import torch
import torch.distributed as dist
import errno
import os
class SmoothedValue(object):
"""Track a series of values and provide access to smoothed values over a
window or the global series average.
"""
def __init__(self, window_size=20, fmt=None):
if fmt is None:
fmt = "{median:.4f} ({global_avg:.4f})"
self.deque = deque(maxlen=window_size)
self.total = 0.0
self.count = 0
self.fmt = fmt
def update(self, value, n=1):
self.deque.append(value)
self.count += n
self.total += value * n
def synchronize_between_processes(self):
"""
Warning: does not synchronize the deque!
"""
if not is_dist_avail_and_initialized():
return
t = torch.tensor([self.count, self.total], dtype=torch.float64, device='cuda')
dist.barrier()
dist.all_reduce(t)
t = t.tolist()
self.count = int(t[0])
self.total = t[1]
@property
def median(self):
d = torch.tensor(list(self.deque))
return d.median().item()
@property
def avg(self):
d = torch.tensor(list(self.deque), dtype=torch.float32)
return d.mean().item()
@property
def global_avg(self):
return self.total / self.count
@property
def max(self):
return max(self.deque)
@property
def value(self):
return self.deque[-1]
def __str__(self):
return self.fmt.format(
median=self.median,
avg=self.avg,
global_avg=self.global_avg,
max=self.max,
value=self.value)
class MetricLogger(object):
def __init__(self, delimiter="\t"):
self.meters = defaultdict(SmoothedValue)
self.delimiter = delimiter
def update(self, **kwargs):
for k, v in kwargs.items():
if isinstance(v, torch.Tensor):
v = v.item()
assert isinstance(v, (float, int))
self.meters[k].update(v)
def __getattr__(self, attr):
if attr in self.meters:
return self.meters[attr]
if attr in self.__dict__:
return self.__dict__[attr]
raise AttributeError("'{}' object has no attribute '{}'".format(
type(self).__name__, attr))
def __str__(self):
loss_str = []
for name, meter in self.meters.items():
loss_str.append(
"{}: {}".format(name, str(meter))
)
return self.delimiter.join(loss_str)
def synchronize_between_processes(self):
for meter in self.meters.values():
meter.synchronize_between_processes()
def add_meter(self, name, meter):
self.meters[name] = meter
def log_every(self, iterable, print_freq, header=None):
i = 0
if not header:
header = ''
start_time = time.time()
end = time.time()
iter_time = SmoothedValue(fmt='{avg:.4f}')
data_time = SmoothedValue(fmt='{avg:.4f}')
space_fmt = ':' + str(len(str(len(iterable)))) + 'd'
if torch.cuda.is_available():
log_msg = self.delimiter.join([
header,
'[{0' + space_fmt + '}/{1}]',
'eta: {eta}',
'{meters}',
'time: {time}',
'data: {data}',
'max mem: {memory:.0f}'
])
else:
log_msg = self.delimiter.join([
header,
'[{0' + space_fmt + '}/{1}]',
'eta: {eta}',
'{meters}',
'time: {time}',
'data: {data}'
])
MB = 1024.0 * 1024.0
for obj in iterable:
data_time.update(time.time() - end)
yield obj
iter_time.update(time.time() - end)
if i % print_freq == 0:
eta_seconds = iter_time.global_avg * (len(iterable) - i)
eta_string = str(datetime.timedelta(seconds=int(eta_seconds)))
if torch.cuda.is_available():
print(log_msg.format(
i, len(iterable), eta=eta_string,
meters=str(self),
time=str(iter_time), data=str(data_time),
memory=torch.cuda.max_memory_allocated() / MB))
else:
print(log_msg.format(
i, len(iterable), eta=eta_string,
meters=str(self),
time=str(iter_time), data=str(data_time)))
i += 1
end = time.time()
total_time = time.time() - start_time
total_time_str = str(datetime.timedelta(seconds=int(total_time)))
print('{} Total time: {}'.format(header, total_time_str))
def accuracy(output, target, topk=(1,)):
"""Computes the accuracy over the k top predictions for the specified values of k"""
with torch.no_grad():
maxk = max(topk)
batch_size = target.size(0)
_, pred = output.topk(maxk, 1, True, True)
pred = pred.t()
correct = pred.eq(target[None])
res = []
for k in topk:
correct_k = correct[:k].flatten().sum(dtype=torch.float32)
res.append(correct_k * (100.0 / batch_size))
return res
def mkdir(path):
try:
os.makedirs(path)
except OSError as e:
if e.errno != errno.EEXIST:
raise
def setup_for_distributed(is_master):
"""
This function disables printing when not in master process
"""
import builtins as __builtin__
builtin_print = __builtin__.print
def print(*args, **kwargs):
force = kwargs.pop('force', False)
if is_master or force:
builtin_print(*args, **kwargs)
__builtin__.print = print
def is_dist_avail_and_initialized():
if not dist.is_available():
return False
if not dist.is_initialized():
return False
return True
def get_world_size():
if not is_dist_avail_and_initialized():
return 1
return dist.get_world_size()
def get_rank():
if not is_dist_avail_and_initialized():
return 0
return dist.get_rank()
def is_main_process():
return get_rank() == 0
def save_on_master(*args, **kwargs):
if is_main_process():
torch.save(*args, **kwargs)
def init_distributed_mode(args):
if 'RANK' in os.environ and 'WORLD_SIZE' in os.environ:
args.rank = int(os.environ["RANK"])
args.world_size = int(os.environ['WORLD_SIZE'])
args.gpu = int(os.environ['LOCAL_RANK'])
elif 'SLURM_PROCID' in os.environ:
args.rank = int(os.environ['SLURM_PROCID'])
args.gpu = args.rank % torch.cuda.device_count()
elif hasattr(args, "rank"):
pass
else:
print('Not using distributed mode')
args.distributed = False
return
args.distributed = True
torch.cuda.set_device(args.gpu)
args.dist_backend = 'nccl'
print('| distributed init (rank {}): {}'.format(
args.rank, args.dist_url), flush=True)
torch.distributed.init_process_group(backend=args.dist_backend, init_method=args.dist_url,
world_size=args.world_size, rank=args.rank)
setup_for_distributed(args.rank == 0)
def average_checkpoints(inputs):
"""Loads checkpoints from inputs and returns a model with averaged weights. Original implementation taken from:
https://github.com/pytorch/fairseq/blob/a48f235636557b8d3bc4922a6fa90f3a0fa57955/scripts/average_checkpoints.py#L16
Args:
inputs (List[str]): An iterable of string paths of checkpoints to load from.
Returns:
A dict of string keys mapping to various values. The 'model' key
from the returned dict should correspond to an OrderedDict mapping
string parameter names to torch Tensors.
"""
params_dict = OrderedDict()
params_keys = None
new_state = None
num_models = len(inputs)
for fpath in inputs:
with open(fpath, "rb") as f:
state = torch.load(
f,
map_location=(
lambda s, _: torch.serialization.default_restore_location(s, "cpu")
),
)
# Copies over the settings from the first checkpoint
if new_state is None:
new_state = state
model_params = state["model"]
model_params_keys = list(model_params.keys())
if params_keys is None:
params_keys = model_params_keys
elif params_keys != model_params_keys:
raise KeyError(
"For checkpoint {}, expected list of params: {}, "
"but found: {}".format(f, params_keys, model_params_keys)
)
for k in params_keys:
p = model_params[k]
if isinstance(p, torch.HalfTensor):
p = p.float()
if k not in params_dict:
params_dict[k] = p.clone()
# NOTE: clone() is needed in case of p is a shared parameter
else:
params_dict[k] += p
averaged_params = OrderedDict()
for k, v in params_dict.items():
averaged_params[k] = v
if averaged_params[k].is_floating_point():
averaged_params[k].div_(num_models)
else:
averaged_params[k] //= num_models
new_state["model"] = averaged_params
return new_state
def store_model_weights(model, checkpoint_path, checkpoint_key='model', strict=True):
"""
This method can be used to prepare weights files for new models. It receives as
input a model architecture and a checkpoint from the training script and produces
a file with the weights ready for release.
Examples:
from torchvision import models as M
# Classification
model = M.mobilenet_v3_large(pretrained=False)
print(store_model_weights(model, './class.pth'))
# Quantized Classification
model = M.quantization.mobilenet_v3_large(pretrained=False, quantize=False)
model.fuse_model()
model.qconfig = torch.quantization.get_default_qat_qconfig('qnnpack')
_ = torch.quantization.prepare_qat(model, inplace=True)
print(store_model_weights(model, './qat.pth'))
# Object Detection
model = M.detection.fasterrcnn_mobilenet_v3_large_fpn(pretrained=False, pretrained_backbone=False)
print(store_model_weights(model, './obj.pth'))
# Segmentation
model = M.segmentation.deeplabv3_mobilenet_v3_large(pretrained=False, pretrained_backbone=False, aux_loss=True)
print(store_model_weights(model, './segm.pth', strict=False))
Args:
model (pytorch.nn.Module): The model on which the weights will be loaded for validation purposes.
checkpoint_path (str): The path of the checkpoint we will load.
checkpoint_key (str, optional): The key of the checkpoint where the model weights are stored.
Default: "model".
strict (bool): whether to strictly enforce that the keys
in :attr:`state_dict` match the keys returned by this module's
:meth:`~torch.nn.Module.state_dict` function. Default: ``True``
Returns:
output_path (str): The location where the weights are saved.
"""
# Store the new model next to the checkpoint_path
checkpoint_path = os.path.abspath(checkpoint_path)
output_dir = os.path.dirname(checkpoint_path)
# Deep copy to avoid side-effects on the model object.
model = copy.deepcopy(model)
checkpoint = torch.load(checkpoint_path, map_location='cpu')
# Load the weights to the model to validate that everything works
# and remove unnecessary weights (such as auxiliaries, etc)
model.load_state_dict(checkpoint[checkpoint_key], strict=strict)
tmp_path = os.path.join(output_dir, str(model.__hash__()))
torch.save(model.state_dict(), tmp_path)
sha256_hash = hashlib.sha256()
with open(tmp_path, "rb") as f:
# Read and update hash string value in blocks of 4K
for byte_block in iter(lambda: f.read(4096), b""):
sha256_hash.update(byte_block)
hh = sha256_hash.hexdigest()
output_path = os.path.join(output_dir, "weights-" + str(hh[:8]) + ".pth")
os.replace(tmp_path, output_path)
return output_path
val.py 0 → 100644
View file @ cea8529d
import datetime
import os
import time
import torch
import torch.utils.data
from torch import nn
import torchvision
import tvm
from tvm.contrib.torch import compile
import presets
import utils
import numpy as np
try:
from apex import amp
except ImportError:
amp = None
def train_one_epoch(model, criterion, optimizer, data_loader, device, epoch, print_freq, apex=False):
model.train()
metric_logger = utils.MetricLogger(delimiter=" ")
metric_logger.add_meter('lr', utils.SmoothedValue(window_size=1, fmt='{value}'))
metric_logger.add_meter('img/s', utils.SmoothedValue(window_size=10, fmt='{value}'))
header = 'Epoch: [{}]'.format(epoch)
for image, target in metric_logger.log_every(data_loader, print_freq, header):
start_time = time.time()
image, target = image.to(device), target.to(device)
output = model(image)
loss = criterion(output, target)
optimizer.zero_grad()
if apex:
with amp.scale_loss(loss, optimizer) as scaled_loss:
scaled_loss.backward()
else:
loss.backward()
optimizer.step()
acc1, acc5 = utils.accuracy(output, target, topk=(1, 5))
batch_size = image.shape[0]
metric_logger.update(loss=loss.item(), lr=optimizer.param_groups[0]["lr"])
metric_logger.meters['acc1'].update(acc1.item(), n=batch_size)
metric_logger.meters['acc5'].update(acc5.item(), n=batch_size)
metric_logger.meters['img/s'].update(batch_size / (time.time() - start_time))
def evaluate(model, criterion, data_loader, device, print_freq=100):
model.eval()
x = torch.rand([1, 3, 224, 224]).cuda()
# 转换为TorchScript格式
scripted_model = torch.jit.trace(model,x)
# 将TorchScript模型保存为.pt文件
torch.jit.save(scripted_model, "model.pt")
model_jit = torch.jit.load("./model.pt")
option = {
"input_infos": [
("x", (1, 3, 224, 224)),
],
"default_dtype": "float32",
"export_dir": "pytorch_compiled",
"num_outputs": 1,
"tuning_n_trials": 0, # set zero to skip tuning
"tuning_log_file": "tuning.log",
"target": "rocm --libs=miopen,rocblas",
"device": tvm.rocm(),
}
pytorch_tvm_module = compile(model_jit, option)
preds = pytorch_tvm_module.forward([x])
preds = pytorch_tvm_module.forward([x])
metric_logger = utils.MetricLogger(delimiter=" ")
header = 'Test:'
with torch.no_grad():
for image, target in metric_logger.log_every(data_loader, print_freq, header):
image = image.to(device, non_blocking=True)
target = target.to(device, non_blocking=True)
output = model(image)
print("print ===orig output =====",output)
output = pytorch_tvm_module.forward([image])
print("print ===output =====",output)
print("print ===========target======",target)
loss = criterion(output, target)
acc1, acc5 = utils.accuracy(output, target, topk=(1, 5))
# FIXME need to take into account that the datasets
# could have been padded in distributed setup
batch_size = image.shape[0]
metric_logger.update(loss=loss.item())
metric_logger.meters['acc1'].update(acc1.item(), n=batch_size)
metric_logger.meters['acc5'].update(acc5.item(), n=batch_size)
# gather the stats from all processes
metric_logger.synchronize_between_processes()
print(' * Acc@1 {top1.global_avg:.3f} Acc@5 {top5.global_avg:.3f}'
.format(top1=metric_logger.acc1, top5=metric_logger.acc5))
return metric_logger.acc1.global_avg
def _get_cache_path(filepath):
import hashlib
h = hashlib.sha1(filepath.encode()).hexdigest()
cache_path = os.path.join("~", ".torch", "vision", "datasets", "imagefolder", h[:10] + ".pt")
cache_path = os.path.expanduser(cache_path)
return cache_path
def load_data(traindir, valdir, args):
# Data loading code
print("Loading data")
resize_size, crop_size = (342, 299) if args.model == 'inception_v3' else (256, 224)
print("Loading training data")
st = time.time()
cache_path = _get_cache_path(traindir)
if args.cache_dataset and os.path.exists(cache_path):
# Attention, as the transforms are also cached!
print("Loading dataset_train from {}".format(cache_path))
dataset, _ = torch.load(cache_path)
else:
auto_augment_policy = getattr(args, "auto_augment", None)
random_erase_prob = getattr(args, "random_erase", 0.0)
dataset = torchvision.datasets.ImageFolder(
traindir,
presets.ClassificationPresetTrain(crop_size=crop_size, auto_augment_policy=auto_augment_policy,
random_erase_prob=random_erase_prob))
if args.cache_dataset:
print("Saving dataset_train to {}".format(cache_path))
utils.mkdir(os.path.dirname(cache_path))
utils.save_on_master((dataset, traindir), cache_path)
print("Took", time.time() - st)
print("Loading validation data")
cache_path = _get_cache_path(valdir)
if args.cache_dataset and os.path.exists(cache_path):
# Attention, as the transforms are also cached!
print("Loading dataset_test from {}".format(cache_path))
dataset_test, _ = torch.load(cache_path)
else:
dataset_test = torchvision.datasets.ImageFolder(
valdir,
presets.ClassificationPresetEval(crop_size=crop_size, resize_size=resize_size))
if args.cache_dataset:
print("Saving dataset_test to {}".format(cache_path))
utils.mkdir(os.path.dirname(cache_path))
utils.save_on_master((dataset_test, valdir), cache_path)
print("Creating data loaders")
if args.distributed:
train_sampler = torch.utils.data.distributed.DistributedSampler(dataset)
test_sampler = torch.utils.data.distributed.DistributedSampler(dataset_test)
else:
train_sampler = torch.utils.data.RandomSampler(dataset)
test_sampler = torch.utils.data.SequentialSampler(dataset_test)
return dataset, dataset_test, train_sampler, test_sampler
def main(args):
if args.apex and amp is None:
raise RuntimeError("Failed to import apex. Please install apex from https://www.github.com/nvidia/apex "
"to enable mixed-precision training.")
if args.output_dir:
utils.mkdir(args.output_dir)
utils.init_distributed_mode(args)
print(args)
device = torch.device(args.device)
torch.backends.cudnn.benchmark = True
train_dir = os.path.join(args.data_path, 'train')
val_dir = os.path.join(args.data_path, 'val')
dataset, dataset_test, train_sampler, test_sampler = load_data(train_dir, val_dir, args)
data_loader = torch.utils.data.DataLoader(
dataset, batch_size=args.batch_size,
sampler=train_sampler, num_workers=args.workers, pin_memory=True)
data_loader_test = torch.utils.data.DataLoader(
dataset_test, batch_size=args.batch_size,
sampler=test_sampler, num_workers=args.workers, pin_memory=True)
print("Creating model")
model = torchvision.models.__dict__[args.model](pretrained=args.pretrained)
model.to(device)
if args.distributed and args.sync_bn:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
criterion = nn.CrossEntropyLoss()
opt_name = args.opt.lower()
if opt_name == 'sgd':
optimizer = torch.optim.SGD(
model.parameters(), lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay)
elif opt_name == 'rmsprop':
optimizer = torch.optim.RMSprop(model.parameters(), lr=args.lr, momentum=args.momentum,
weight_decay=args.weight_decay, eps=0.0316, alpha=0.9)
else:
raise RuntimeError("Invalid optimizer {}. Only SGD and RMSprop are supported.".format(args.opt))
if args.apex:
model, optimizer = amp.initialize(model, optimizer,
opt_level=args.apex_opt_level
)
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=args.lr_step_size, gamma=args.lr_gamma)
model_without_ddp = model
if args.distributed:
model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu])
model_without_ddp = model.module
if args.resume:
checkpoint = torch.load(args.resume, map_location='cpu')
model_without_ddp.load_state_dict(checkpoint['model'])
optimizer.load_state_dict(checkpoint['optimizer'])
lr_scheduler.load_state_dict(checkpoint['lr_scheduler'])
args.start_epoch = checkpoint['epoch'] + 1
if args.test_only:
evaluate(model, criterion, data_loader_test, device=device)
return
print("Start training")
start_time = time.time()
for epoch in range(args.start_epoch, args.epochs):
if args.distributed:
train_sampler.set_epoch(epoch)
train_one_epoch(model, criterion, optimizer, data_loader, device, epoch, args.print_freq, args.apex)
lr_scheduler.step()
evaluate(model, criterion, data_loader_test, device=device)
if args.output_dir:
checkpoint = {
'model': model_without_ddp.state_dict(),
'optimizer': optimizer.state_dict(),
'lr_scheduler': lr_scheduler.state_dict(),
'epoch': epoch,
'args': args}
utils.save_on_master(
checkpoint,
os.path.join(args.output_dir, 'model_{}.pth'.format(epoch)))
utils.save_on_master(
checkpoint,
os.path.join(args.output_dir, 'checkpoint.pth'))
total_time = time.time() - start_time
total_time_str = str(datetime.timedelta(seconds=int(total_time)))
print('Training time {}'.format(total_time_str))
def get_args_parser(add_help=True):
import argparse
parser = argparse.ArgumentParser(description='PyTorch Classification Training', add_help=add_help)
parser.add_argument('--data-path', default='/datasets01/imagenet_full_size/061417/', help='dataset')
parser.add_argument('--model', default='resnet18', help='model')
parser.add_argument('--device', default='cuda', help='device')
parser.add_argument('-b', '--batch-size', default=32, type=int)
parser.add_argument('--epochs', default=90, type=int, metavar='N',
help='number of total epochs to run')
parser.add_argument('-j', '--workers', default=16, type=int, metavar='N',
help='number of data loading workers (default: 16)')
parser.add_argument('--opt', default='sgd', type=str, help='optimizer')
parser.add_argument('--lr', default=0.1, type=float, help='initial learning rate')
parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
help='momentum')
parser.add_argument('--wd', '--weight-decay', default=1e-4, type=float,
metavar='W', help='weight decay (default: 1e-4)',
dest='weight_decay')
parser.add_argument('--lr-step-size', default=30, type=int, help='decrease lr every step-size epochs')
parser.add_argument('--lr-gamma', default=0.1, type=float, help='decrease lr by a factor of lr-gamma')
parser.add_argument('--print-freq', default=10, type=int, help='print frequency')
parser.add_argument('--output-dir', default='.', help='path where to save')
parser.add_argument('--resume', default='', help='resume from checkpoint')
parser.add_argument('--start-epoch', default=0, type=int, metavar='N',
help='start epoch')
parser.add_argument(
"--cache-dataset",
dest="cache_dataset",
help="Cache the datasets for quicker initialization. It also serializes the transforms",
action="store_true",
)
parser.add_argument(
"--sync-bn",
dest="sync_bn",
help="Use sync batch norm",
action="store_true",
)
parser.add_argument(
"--test-only",
dest="test_only",
help="Only test the model",
action="store_true",
)
parser.add_argument(
"--pretrained",
dest="pretrained",
help="Use pre-trained models from the modelzoo",
action="store_true",
)
parser.add_argument('--auto-augment', default=None, help='auto augment policy (default: None)')
parser.add_argument('--random-erase', default=0.0, type=float, help='random erasing probability (default: 0.0)')
# Mixed precision training parameters
parser.add_argument('--apex', action='store_true',
help='Use apex for mixed precision training')
parser.add_argument('--apex-opt-level', default='O1', type=str,
help='For apex mixed precision training'
'O0 for FP32 training, O1 for mixed precision training.'
'For further detail, see https://github.com/NVIDIA/apex/tree/master/examples/imagenet'
)
# distributed training parameters
parser.add_argument('--world-size', default=1, type=int,
help='number of distributed processes')
parser.add_argument('--dist-url', default='env://', help='url used to set up distributed training')
return parser
if __name__ == "__main__":
args = get_args_parser().parse_args()
main(args)
val.py.bak 0 → 100644
View file @ cea8529d
import datetime
import os
import time
import torch
import torch.utils.data
from torch import nn
import torchvision
import tvm
from tvm.contrib.torch import compile
import presets
import utils
import numpy as np
try:
from apex import amp
except ImportError:
amp = None
def train_one_epoch(model, criterion, optimizer, data_loader, device, epoch, print_freq, apex=False):
model.train()
metric_logger = utils.MetricLogger(delimiter=" ")
metric_logger.add_meter('lr', utils.SmoothedValue(window_size=1, fmt='{value}'))
metric_logger.add_meter('img/s', utils.SmoothedValue(window_size=10, fmt='{value}'))
header = 'Epoch: [{}]'.format(epoch)
for image, target in metric_logger.log_every(data_loader, print_freq, header):
start_time = time.time()
image, target = image.to(device), target.to(device)
output = model(image)
loss = criterion(output, target)
optimizer.zero_grad()
if apex:
with amp.scale_loss(loss, optimizer) as scaled_loss:
scaled_loss.backward()
else:
loss.backward()
optimizer.step()
acc1, acc5 = utils.accuracy(output, target, topk=(1, 5))
batch_size = image.shape[0]
metric_logger.update(loss=loss.item(), lr=optimizer.param_groups[0]["lr"])
metric_logger.meters['acc1'].update(acc1.item(), n=batch_size)
metric_logger.meters['acc5'].update(acc5.item(), n=batch_size)
metric_logger.meters['img/s'].update(batch_size / (time.time() - start_time))
def evaluate(model, criterion, data_loader, device, print_freq=100):
model.eval()
x = torch.rand([1, 3, 224, 224]).cuda()
# 转换为TorchScript格式
scripted_model = torch.jit.trace(model,x)
# 将TorchScript模型保存为.pt文件
torch.jit.save(scripted_model, "model.pt")
model_jit = torch.jit.load("./model.pt")
option = {
"input_infos": [
("x", (1, 3, 224, 224)),
],
"default_dtype": "float32",
"export_dir": "pytorch_compiled",
"num_outputs": 1,
"tuning_n_trials": 0, # set zero to skip tuning
"tuning_log_file": "tuning.log",
"target": "rocm --libs=miopen,rocblas",
"device": tvm.rocm(),
}
pytorch_tvm_module = compile(model_jit, option)
preds = pytorch_tvm_module.forward([x])
preds = pytorch_tvm_module.forward([x])
metric_logger = utils.MetricLogger(delimiter=" ")
header = 'Test:'
with torch.no_grad():
for image, target in metric_logger.log_every(data_loader, print_freq, header):
image = image.to(device, non_blocking=True)
target = target.to(device, non_blocking=True)
output = model(image)
print("print ===orig output =====",output)
output = pytorch_tvm_module.forward([image])
print("print ===output =====",output)
print("print ===========target======",target)
loss = criterion(output, target)
acc1, acc5 = utils.accuracy(output, target, topk=(1, 5))
# FIXME need to take into account that the datasets
# could have been padded in distributed setup
batch_size = image.shape[0]
metric_logger.update(loss=loss.item())
metric_logger.meters['acc1'].update(acc1.item(), n=batch_size)
metric_logger.meters['acc5'].update(acc5.item(), n=batch_size)
# gather the stats from all processes
metric_logger.synchronize_between_processes()
print(' * Acc@1 {top1.global_avg:.3f} Acc@5 {top5.global_avg:.3f}'
.format(top1=metric_logger.acc1, top5=metric_logger.acc5))
return metric_logger.acc1.global_avg
def _get_cache_path(filepath):
import hashlib
h = hashlib.sha1(filepath.encode()).hexdigest()
cache_path = os.path.join("~", ".torch", "vision", "datasets", "imagefolder", h[:10] + ".pt")
cache_path = os.path.expanduser(cache_path)
return cache_path
def load_data(traindir, valdir, args):
# Data loading code
print("Loading data")
resize_size, crop_size = (342, 299) if args.model == 'inception_v3' else (256, 224)
print("Loading training data")
st = time.time()
cache_path = _get_cache_path(traindir)
if args.cache_dataset and os.path.exists(cache_path):
# Attention, as the transforms are also cached!
print("Loading dataset_train from {}".format(cache_path))
dataset, _ = torch.load(cache_path)
else:
auto_augment_policy = getattr(args, "auto_augment", None)
random_erase_prob = getattr(args, "random_erase", 0.0)
dataset = torchvision.datasets.ImageFolder(
traindir,
presets.ClassificationPresetTrain(crop_size=crop_size, auto_augment_policy=auto_augment_policy,
random_erase_prob=random_erase_prob))
if args.cache_dataset:
print("Saving dataset_train to {}".format(cache_path))
utils.mkdir(os.path.dirname(cache_path))
utils.save_on_master((dataset, traindir), cache_path)
print("Took", time.time() - st)
print("Loading validation data")
cache_path = _get_cache_path(valdir)
if args.cache_dataset and os.path.exists(cache_path):
# Attention, as the transforms are also cached!
print("Loading dataset_test from {}".format(cache_path))
dataset_test, _ = torch.load(cache_path)
else:
dataset_test = torchvision.datasets.ImageFolder(
valdir,
presets.ClassificationPresetEval(crop_size=crop_size, resize_size=resize_size))
if args.cache_dataset:
print("Saving dataset_test to {}".format(cache_path))
utils.mkdir(os.path.dirname(cache_path))
utils.save_on_master((dataset_test, valdir), cache_path)
print("Creating data loaders")
if args.distributed:
train_sampler = torch.utils.data.distributed.DistributedSampler(dataset)
test_sampler = torch.utils.data.distributed.DistributedSampler(dataset_test)
else:
train_sampler = torch.utils.data.RandomSampler(dataset)
test_sampler = torch.utils.data.SequentialSampler(dataset_test)
return dataset, dataset_test, train_sampler, test_sampler
def main(args):
if args.apex and amp is None:
raise RuntimeError("Failed to import apex. Please install apex from https://www.github.com/nvidia/apex "
"to enable mixed-precision training.")
if args.output_dir:
utils.mkdir(args.output_dir)
utils.init_distributed_mode(args)
print(args)
device = torch.device(args.device)
torch.backends.cudnn.benchmark = True
train_dir = os.path.join(args.data_path, 'train')
val_dir = os.path.join(args.data_path, 'val')
dataset, dataset_test, train_sampler, test_sampler = load_data(train_dir, val_dir, args)
data_loader = torch.utils.data.DataLoader(
dataset, batch_size=args.batch_size,
sampler=train_sampler, num_workers=args.workers, pin_memory=True)
data_loader_test = torch.utils.data.DataLoader(
dataset_test, batch_size=args.batch_size,
sampler=test_sampler, num_workers=args.workers, pin_memory=True)
print("Creating model")
model = torchvision.models.__dict__[args.model](pretrained=args.pretrained)
model.to(device)
if args.distributed and args.sync_bn:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
criterion = nn.CrossEntropyLoss()
opt_name = args.opt.lower()
if opt_name == 'sgd':
optimizer = torch.optim.SGD(
model.parameters(), lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay)
elif opt_name == 'rmsprop':
optimizer = torch.optim.RMSprop(model.parameters(), lr=args.lr, momentum=args.momentum,
weight_decay=args.weight_decay, eps=0.0316, alpha=0.9)
else:
raise RuntimeError("Invalid optimizer {}. Only SGD and RMSprop are supported.".format(args.opt))
if args.apex:
model, optimizer = amp.initialize(model, optimizer,
opt_level=args.apex_opt_level
)
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=args.lr_step_size, gamma=args.lr_gamma)
model_without_ddp = model
if args.distributed:
model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu])
model_without_ddp = model.module
if args.resume:
checkpoint = torch.load(args.resume, map_location='cpu')
model_without_ddp.load_state_dict(checkpoint['model'])
optimizer.load_state_dict(checkpoint['optimizer'])
lr_scheduler.load_state_dict(checkpoint['lr_scheduler'])
args.start_epoch = checkpoint['epoch'] + 1
if args.test_only:
evaluate(model, criterion, data_loader_test, device=device)
return
print("Start training")
start_time = time.time()
for epoch in range(args.start_epoch, args.epochs):
if args.distributed:
train_sampler.set_epoch(epoch)
train_one_epoch(model, criterion, optimizer, data_loader, device, epoch, args.print_freq, args.apex)
lr_scheduler.step()
evaluate(model, criterion, data_loader_test, device=device)
if args.output_dir:
checkpoint = {
'model': model_without_ddp.state_dict(),
'optimizer': optimizer.state_dict(),
'lr_scheduler': lr_scheduler.state_dict(),
'epoch': epoch,
'args': args}
utils.save_on_master(
checkpoint,
os.path.join(args.output_dir, 'model_{}.pth'.format(epoch)))
utils.save_on_master(
checkpoint,
os.path.join(args.output_dir, 'checkpoint.pth'))
total_time = time.time() - start_time
total_time_str = str(datetime.timedelta(seconds=int(total_time)))
print('Training time {}'.format(total_time_str))
def get_args_parser(add_help=True):
import argparse
parser = argparse.ArgumentParser(description='PyTorch Classification Training', add_help=add_help)
parser.add_argument('--data-path', default='/datasets01/imagenet_full_size/061417/', help='dataset')
parser.add_argument('--model', default='resnet18', help='model')
parser.add_argument('--device', default='cuda', help='device')
parser.add_argument('-b', '--batch-size', default=32, type=int)
parser.add_argument('--epochs', default=90, type=int, metavar='N',
help='number of total epochs to run')
parser.add_argument('-j', '--workers', default=16, type=int, metavar='N',
help='number of data loading workers (default: 16)')
parser.add_argument('--opt', default='sgd', type=str, help='optimizer')
parser.add_argument('--lr', default=0.1, type=float, help='initial learning rate')
parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
help='momentum')
parser.add_argument('--wd', '--weight-decay', default=1e-4, type=float,
metavar='W', help='weight decay (default: 1e-4)',
dest='weight_decay')
parser.add_argument('--lr-step-size', default=30, type=int, help='decrease lr every step-size epochs')
parser.add_argument('--lr-gamma', default=0.1, type=float, help='decrease lr by a factor of lr-gamma')
parser.add_argument('--print-freq', default=10, type=int, help='print frequency')
parser.add_argument('--output-dir', default='.', help='path where to save')
parser.add_argument('--resume', default='', help='resume from checkpoint')
parser.add_argument('--start-epoch', default=0, type=int, metavar='N',
help='start epoch')
parser.add_argument(
"--cache-dataset",
dest="cache_dataset",
help="Cache the datasets for quicker initialization. It also serializes the transforms",
action="store_true",
)
parser.add_argument(
"--sync-bn",
dest="sync_bn",
help="Use sync batch norm",
action="store_true",
)
parser.add_argument(
"--test-only",
dest="test_only",
help="Only test the model",
action="store_true",
)
parser.add_argument(
"--pretrained",
dest="pretrained",
help="Use pre-trained models from the modelzoo",
action="store_true",
)
parser.add_argument('--auto-augment', default=None, help='auto augment policy (default: None)')
parser.add_argument('--random-erase', default=0.0, type=float, help='random erasing probability (default: 0.0)')
# Mixed precision training parameters
parser.add_argument('--apex', action='store_true',
help='Use apex for mixed precision training')
parser.add_argument('--apex-opt-level', default='O1', type=str,
help='For apex mixed precision training'
'O0 for FP32 training, O1 for mixed precision training.'
'For further detail, see https://github.com/NVIDIA/apex/tree/master/examples/imagenet'
)
# distributed training parameters
parser.add_argument('--world-size', default=1, type=int,
help='number of distributed processes')
parser.add_argument('--dist-url', default='env://', help='url used to set up distributed training')
return parser
if __name__ == "__main__":
args = get_args_parser().parse_args()
main(args)
val_onnx.py 0 → 100644
View file @ cea8529d
import datetime
import os
import time
import numpy as np
import onnx
import torch
import torch.utils.data
from torch import nn
import torchvision
import tvm
from tvm.contrib.torch import compile
from tvm import relay
from tvm.contrib import graph_executor
import presets
import utils
try:
from apex import amp
except ImportError:
amp = None
def train_one_epoch(model, criterion, optimizer, data_loader, device, epoch, print_freq, apex=False):
model.train()
metric_logger = utils.MetricLogger(delimiter=" ")
metric_logger.add_meter('lr', utils.SmoothedValue(window_size=1, fmt='{value}'))
#metric_logger.add_meter('img/s', utils.SmoothedValue(window_size=10, fmt='{value}'))
header = 'Epoch: [{}]'.format(epoch)
for image, target in metric_logger.log_every(data_loader, print_freq, header):
start_time = time.time()
image, target = image.to(device), target.to(device)
output = model(image)
loss = criterion(output, target)
optimizer.zero_grad()
if apex:
with amp.scale_loss(loss, optimizer) as scaled_loss:
scaled_loss.backward()
else:
loss.backward()
optimizer.step()
acc1, acc5 = utils.accuracy(output, target, topk=(1, 5))
batch_size = image.shape[0]
metric_logger.update(loss=loss.item(), lr=optimizer.param_groups[0]["lr"])
metric_logger.meters['acc1'].update(acc1.item(), n=batch_size)
metric_logger.meters['acc5'].update(acc5.item(), n=batch_size)
metric_logger.meters['img/s'].update(batch_size / (time.time() - start_time))
def evaluate(model, criterion, data_loader, device, print_freq=100):
model.eval()
# 转换为TorchScript格式
input_shape = (1, 3, 224, 224)
input_tensor = torch.randn(input_shape).to(torch.device("cuda"))
# 导出ONNX文件
output_path = "./mobilenet.onnx"
torch.onnx.export(model, input_tensor, output_path)
# 将TorchScript模型保存为.pt文件
img_data = np.random.rand(1,3,224,224).astype("float32")/255
input_name = "input.1"
shape_dict = {input_name: img_data.shape}
model_path = './mobilenet.onnx'
onnx_model = onnx.load(model_path)
# Define the neural network and compilation target
batch_size = 1
layout = "NCHW"
target = "rocm -libs=miopen,rocblas"
dtype = "float32"
mod, params = relay.frontend.from_onnx(onnx_model, shape_dict, dtype=dtype)
# Compile with the history best
print("Compile...")
#mod = tvm.relay.transform.ToMixedPrecision(mixed_precision_type='float16', missing_op_mode=1)(mod)
with tvm.transform.PassContext(opt_level=3):
lib = relay.build(mod, target=target, params=params)
print('Compile success!')
dev = tvm.device(str(target), 0)
m = graph_executor.GraphModule(lib["default"](dev))
metric_logger = utils.MetricLogger(delimiter=" ")
header = 'Test:'
with torch.no_grad():
for image, target in metric_logger.log_every(data_loader, print_freq, header):
# image = image.to(device, non_blocking=True)
target = target.to(device, non_blocking=True)
# output = model(image)
m.set_input("input.1", image.numpy())
m.run()
out = m.get_output(0).asnumpy()
output = torch.from_numpy(out).to(torch.device("cuda"))
loss = criterion(output, target)
acc1, acc5 = utils.accuracy(output, target, topk=(1, 5))
# FIXME need to take into account that the datasets
# could have been padded in distributed setup
batch_size = image.shape[0]
metric_logger.update(loss=loss.item())
metric_logger.meters['acc1'].update(acc1.item(), n=batch_size)
metric_logger.meters['acc5'].update(acc5.item(), n=batch_size)
# gather the stats from all processes
metric_logger.synchronize_between_processes()
print(' * Acc@1 {top1.global_avg:.3f} Acc@5 {top5.global_avg:.3f}'
.format(top1=metric_logger.acc1, top5=metric_logger.acc5))
return metric_logger.acc1.global_avg
def _get_cache_path(filepath):
import hashlib
h = hashlib.sha1(filepath.encode()).hexdigest()
cache_path = os.path.join("~", ".torch", "vision", "datasets", "imagefolder", h[:10] + ".pt")
cache_path = os.path.expanduser(cache_path)
return cache_path
def load_data(traindir, valdir, args):
# Data loading code
print("Loading data")
resize_size, crop_size = (342, 299) if args.model == 'inception_v3' else (256, 224)
print("Loading training data")
st = time.time()
cache_path = _get_cache_path(traindir)
if args.cache_dataset and os.path.exists(cache_path):
# Attention, as the transforms are also cached!
print("Loading dataset_train from {}".format(cache_path))
dataset, _ = torch.load(cache_path)
else:
auto_augment_policy = getattr(args, "auto_augment", None)
random_erase_prob = getattr(args, "random_erase", 0.0)
dataset = torchvision.datasets.ImageFolder(
traindir,
presets.ClassificationPresetTrain(crop_size=crop_size, auto_augment_policy=auto_augment_policy,
random_erase_prob=random_erase_prob))
if args.cache_dataset:
print("Saving dataset_train to {}".format(cache_path))
utils.mkdir(os.path.dirname(cache_path))
utils.save_on_master((dataset, traindir), cache_path)
print("Took", time.time() - st)
print("Loading validation data")
cache_path = _get_cache_path(valdir)
if args.cache_dataset and os.path.exists(cache_path):
# Attention, as the transforms are also cached!
print("Loading dataset_test from {}".format(cache_path))
dataset_test, _ = torch.load(cache_path)
else:
dataset_test = torchvision.datasets.ImageFolder(
valdir,
presets.ClassificationPresetEval(crop_size=crop_size, resize_size=resize_size))
if args.cache_dataset:
print("Saving dataset_test to {}".format(cache_path))
utils.mkdir(os.path.dirname(cache_path))
utils.save_on_master((dataset_test, valdir), cache_path)
print("Creating data loaders")
if args.distributed:
train_sampler = torch.utils.data.distributed.DistributedSampler(dataset)
test_sampler = torch.utils.data.distributed.DistributedSampler(dataset_test)
else:
train_sampler = torch.utils.data.RandomSampler(dataset)
test_sampler = torch.utils.data.SequentialSampler(dataset_test)
return dataset, dataset_test, train_sampler, test_sampler
def main(args):
if args.apex and amp is None:
raise RuntimeError("Failed to import apex. Please install apex from https://www.github.com/nvidia/apex "
"to enable mixed-precision training.")
if args.output_dir:
utils.mkdir(args.output_dir)
utils.init_distributed_mode(args)
print(args)
device = torch.device(args.device)
torch.backends.cudnn.benchmark = True
train_dir = os.path.join(args.data_path, 'train')
val_dir = os.path.join(args.data_path, 'val')
dataset, dataset_test, train_sampler, test_sampler = load_data(train_dir, val_dir, args)
data_loader = torch.utils.data.DataLoader(
dataset, batch_size=args.batch_size,
sampler=train_sampler, num_workers=args.workers, pin_memory=True)
data_loader_test = torch.utils.data.DataLoader(
dataset_test, batch_size=args.batch_size,
sampler=test_sampler, num_workers=args.workers, pin_memory=True)
print("Creating model")
model = torchvision.models.__dict__[args.model](pretrained=args.pretrained)
model.to(device)
if args.distributed and args.sync_bn:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
criterion = nn.CrossEntropyLoss()
opt_name = args.opt.lower()
if opt_name == 'sgd':
optimizer = torch.optim.SGD(
model.parameters(), lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay)
elif opt_name == 'rmsprop':
optimizer = torch.optim.RMSprop(model.parameters(), lr=args.lr, momentum=args.momentum,
weight_decay=args.weight_decay, eps=0.0316, alpha=0.9)
else:
raise RuntimeError("Invalid optimizer {}. Only SGD and RMSprop are supported.".format(args.opt))
if args.apex:
model, optimizer = amp.initialize(model, optimizer,
opt_level=args.apex_opt_level
)
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=args.lr_step_size, gamma=args.lr_gamma)
model_without_ddp = model
if args.distributed:
model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu])
model_without_ddp = model.module
if args.resume:
checkpoint = torch.load(args.resume, map_location='cpu')
model_without_ddp.load_state_dict(checkpoint['model'])
optimizer.load_state_dict(checkpoint['optimizer'])
lr_scheduler.load_state_dict(checkpoint['lr_scheduler'])
args.start_epoch = checkpoint['epoch'] + 1
if args.test_only:
evaluate(model, criterion, data_loader_test, device=device)
return
print("Start training")
start_time = time.time()
for epoch in range(args.start_epoch, args.epochs):
if args.distributed:
train_sampler.set_epoch(epoch)
train_one_epoch(model, criterion, optimizer, data_loader, device, epoch, args.print_freq, args.apex)
lr_scheduler.step()
evaluate(model, criterion, data_loader_test, device=device)
if args.output_dir:
checkpoint = {
'model': model_without_ddp.state_dict(),
'optimizer': optimizer.state_dict(),
'lr_scheduler': lr_scheduler.state_dict(),
'epoch': epoch,
'args': args}
utils.save_on_master(
checkpoint,
os.path.join(args.output_dir, 'model_{}.pth'.format(epoch)))
utils.save_on_master(
checkpoint,
os.path.join(args.output_dir, 'checkpoint.pth'))
total_time = time.time() - start_time
total_time_str = str(datetime.timedelta(seconds=int(total_time)))
print('Training time {}'.format(total_time_str))
def get_args_parser(add_help=True):
import argparse
parser = argparse.ArgumentParser(description='PyTorch Classification Training', add_help=add_help)
parser.add_argument('--data-path', default='/datasets01/imagenet_full_size/061417/', help='dataset')
parser.add_argument('--model', default='resnet18', help='model')
parser.add_argument('--device', default='cuda', help='device')
parser.add_argument('-b', '--batch-size', default=32, type=int)
parser.add_argument('--epochs', default=90, type=int, metavar='N',
help='number of total epochs to run')
parser.add_argument('-j', '--workers', default=16, type=int, metavar='N',
help='number of data loading workers (default: 16)')
parser.add_argument('--opt', default='sgd', type=str, help='optimizer')
parser.add_argument('--lr', default=0.1, type=float, help='initial learning rate')
parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
help='momentum')
parser.add_argument('--wd', '--weight-decay', default=1e-4, type=float,
metavar='W', help='weight decay (default: 1e-4)',
dest='weight_decay')
parser.add_argument('--lr-step-size', default=30, type=int, help='decrease lr every step-size epochs')
parser.add_argument('--lr-gamma', default=0.1, type=float, help='decrease lr by a factor of lr-gamma')
parser.add_argument('--print-freq', default=10, type=int, help='print frequency')
parser.add_argument('--output-dir', default='.', help='path where to save')
parser.add_argument('--resume', default='', help='resume from checkpoint')
parser.add_argument('--start-epoch', default=0, type=int, metavar='N',
help='start epoch')
parser.add_argument(
"--cache-dataset",
dest="cache_dataset",
help="Cache the datasets for quicker initialization. It also serializes the transforms",
action="store_true",
)
parser.add_argument(
"--sync-bn",
dest="sync_bn",
help="Use sync batch norm",
action="store_true",
)
parser.add_argument(
"--test-only",
dest="test_only",
help="Only test the model",
action="store_true",
)
parser.add_argument(
"--pretrained",
dest="pretrained",
help="Use pre-trained models from the modelzoo",
action="store_true",
)
parser.add_argument('--auto-augment', default=None, help='auto augment policy (default: None)')
parser.add_argument('--random-erase', default=0.0, type=float, help='random erasing probability (default: 0.0)')
# Mixed precision training parameters
parser.add_argument('--apex', action='store_true',
help='Use apex for mixed precision training')
parser.add_argument('--apex-opt-level', default='O1', type=str,
help='For apex mixed precision training'
'O0 for FP32 training, O1 for mixed precision training.'
'For further detail, see https://github.com/NVIDIA/apex/tree/master/examples/imagenet'
)
# distributed training parameters
parser.add_argument('--world-size', default=1, type=int,
help='number of distributed processes')
parser.add_argument('--dist-url', default='env://', help='url used to set up distributed training')
return parser
if __name__ == "__main__":
args = get_args_parser().parse_args()
main(args)
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment