Pix2Pix实现图像转换

tutorials/application/source_zh_cn/generative/pix2pix.ipynb · MindSpore/docs - Gitee.com

Pix2Pix概述

Pix2Pix是基于条件生成对抗网络(cGAN, Condition Generative Adversarial Networks )实现的一种深度学习图像转换模型,该模型是由Phillip Isola等作者在2017年CVPR上提出的,可以实现语义/标签到真实图片、灰度图到彩色图、航空图到地图、白天到黑夜、线稿图到实物图的转换。Pix2Pix是将cGAN应用于有监督的图像到图像翻译的经典之作,其包括两个模型:生成器判别器

传统上,尽管此类任务的目标都是相同的从像素预测像素,但每项都是用单独的专用机器来处理的。而Pix2Pix使用的网络作为一个通用框架,使用相同的架构和目标,只在不同的数据上进行训练,即可得到令人满意的结果,鉴于此许多人已经使用此网络发布了他们自己的艺术作品。

基础原理

cGAN的生成器与传统GAN的生成器在原理上有一些区别,cGAN的生成器是将输入图片作为指导信息,由输入图像不断尝试生成用于迷惑判别器的“假”图像,由输入图像转换输出为相应“假”图像的本质是从像素到另一个像素的映射,而传统GAN的生成器是基于一个给定的随机噪声生成图像,输出图像通过其他约束条件控制生成,这是cGAN和GAN的在图像翻译任务中的差异。Pix2Pix中判别器的任务是判断从生成器输出的图像是真实的训练图像还是生成的“假”图像。在生成器与判别器的不断博弈过程中,模型会达到一个平衡点,生成器输出的图像与真实训练数据使得判别器刚好具有50%的概率判断正确。

在教程开始前,首先定义一些在整个过程中需要用到的符号:

  • 𝑥:代表观测图像的数据。
  • 𝑧:代表随机噪声的数据。
  • 𝑦=𝐺(𝑥,𝑧):生成器网络,给出由观测图像𝑥与随机噪声𝑧生成的“假”图片,其中𝑥来自于训练数据而非生成器。
  • 𝐷(𝑥,𝐺(𝑥,𝑧)):判别器网络,给出图像判定为真实图像的概率,其中𝑥来自于训练数据,𝐺(𝑥,𝑧)来自于生成器。

cGAN的目标可以表示为:

𝐿𝑐𝐺𝐴𝑁(𝐺,𝐷)=𝐸(𝑥,𝑦)[𝑙𝑜𝑔(𝐷(𝑥,𝑦))]+𝐸(𝑥,𝑧)[𝑙𝑜𝑔(1−𝐷(𝑥,𝐺(𝑥,𝑧)))]

该公式是cGAN的损失函数,D想要尽最大努力去正确分类真实图像与“假”图像,也就是使参数𝑙𝑜𝑔𝐷(𝑥,𝑦)最大化;而G则尽最大努力用生成的“假”图像𝑦欺骗D,避免被识破,也就是使参数𝑙𝑜𝑔(1−𝐷(𝐺(𝑥,𝑧)))最小化。cGAN的目标可简化为:

arg \underset{G}{min} \underset{D}{max} L_{cGAN}(G,D)

pix2pix1

为了对比cGAN和GAN的不同,我们将GAN的目标也进行了说明:

𝐿𝐺𝐴𝑁(𝐺,𝐷)=𝐸𝑦[𝑙𝑜𝑔(𝐷(𝑦))]+𝐸(𝑥,𝑧)[𝑙𝑜𝑔(1−𝐷(𝑥,𝑧))]

从公式可以看出,GAN直接由随机噪声𝑧z生成“假”图像,不借助观测图像𝑥x的任何信息。过去的经验告诉我们,GAN与传统损失混合使用是有好处的,判别器的任务不变,依旧是区分真实图像与“假”图像,但是生成器的任务不仅要欺骗判别器,还要在传统损失的基础上接近训练数据。假设cGAN与L1正则化混合使用,那么有:

𝐿𝐿1(𝐺)=𝐸(𝑥,𝑦,𝑧)[||𝑦−𝐺(𝑥,𝑧)||1]

进而得到最终目标:

𝑎𝑟𝑔min𝐺max𝐷𝐿𝑐𝐺𝐴𝑁(𝐺,𝐷)+𝜆𝐿𝐿1(𝐺)

图像转换问题本质上其实就是像素到像素的映射问题,Pix2Pix使用完全一样的网络结构和目标函数,仅更换不同的训练数据集就能分别实现以上的任务。本任务将借助MindSpore框架来实现Pix2Pix的应用。

准备环节

配置环境文件

本案例在GPU,CPU和Ascend平台的动静态模式都支持。

准备数据

在本教程中,我们将使用指定数据集,该数据集是已经经过处理的外墙(facades)数据,可以直接使用mindspore.dataset的方法读取。

%%capture captured_output
# 实验环境已经预装了mindspore==2.3.0,如需更换mindspore版本,可更改下面 MINDSPORE_VERSION 变量
!pip uninstall mindspore -y
%env MINDSPORE_VERSION=2.3.0
!pip install https://ms-release.obs.cn-north-4.myhuaweicloud.com/${MINDSPORE_VERSION}/MindSpore/unified/aarch64/mindspore-${MINDSPORE_VERSION}-cp39-cp39-linux_aarch64.whl --trusted-host ms-release.obs.cn-north-4.myhuaweicloud.com -i https://pypi.mirrors.ustc.edu.cn/simple
# 查看当前 mindspore 版本
!pip show mindspore
Name: mindspore
Version: 2.3.0
Summary: MindSpore is a new open source deep learning training/inference framework that could be used for mobile, edge and cloud scenarios.
Home-page: https://www.mindspore.cn
Author: The MindSpore Authors
Author-email: contact@mindspore.cn
License: Apache 2.0
Location: /home/mindspore/miniconda/envs/jupyter/lib/python3.9/site-packages
Requires: asttokens, astunparse, numpy, packaging, pillow, protobuf, psutil, scipy
Required-by: 
from download import downloadurl = "https://mindspore-website.obs.cn-north-4.myhuaweicloud.com/notebook/models/application/dataset_pix2pix.tar"download(url, "./dataset", kind="tar", replace=True)
Downloading data from https://mindspore-website.obs.cn-north-4.myhuaweicloud.com/notebook/models/application/dataset_pix2pix.tar (840.0 MB)file_sizes: 100%|█████████████████████████████| 881M/881M [00:05<00:00, 175MB/s]
Extracting tar file...
Successfully downloaded / unzipped to ./dataset

[3]:

'./dataset'

数据展示

调用Pix2PixDatasetcreate_train_dataset读取训练集,这里我们直接下载已经处理好的数据集。

from mindspore import dataset as ds
import matplotlib.pyplot as pltdataset = ds.MindDataset("./dataset/dataset_pix2pix/train.mindrecord", columns_list=["input_images", "target_images"], shuffle=True)
data_iter = next(dataset.create_dict_iterator(output_numpy=True))
# 可视化部分训练数据
plt.figure(figsize=(10, 3), dpi=140)
for i, image in enumerate(data_iter['input_images'][:10], 1):plt.subplot(3, 10, i)plt.axis("off")plt.imshow((image.transpose(1, 2, 0) + 1) / 2)
plt.show()

创建网络

当处理完数据后,就可以来进行网络的搭建了。网络搭建将逐一详细讨论生成器、判别器和损失函数。生成器G用到的是U-Net结构,输入的轮廓图𝑥x编码再解码成真是图片,判别器D用到的是作者自己提出来的条件判别器PatchGAN,判别器D的作用是在轮廓图 𝑥x的条件下,对于生成的图片𝐺(𝑥)G(x)判断为假,对于真实判断为真。

生成器G结构

U-Net是德国Freiburg大学模式识别和图像处理组提出的一种全卷积结构。它分为两个部分,其中左侧是由卷积和降采样操作组成的压缩路径,右侧是由卷积和上采样组成的扩张路径,扩张的每个网络块的输入由上一层上采样的特征和压缩路径部分的特征拼接而成。网络模型整体是一个U形的结构,因此被叫做U-Net。和常见的先降采样到低维度,再升采样到原始分辨率的编解码结构的网络相比,U-Net的区别是加入skip-connection,对应的feature maps和decode之后的同样大小的feature maps按通道拼一起,用来保留不同分辨率下像素级的细节信息。

pix2pix2

定义UNet Skip Connection Block
import mindspore
import mindspore.nn as nn
import mindspore.ops as opsclass UNetSkipConnectionBlock(nn.Cell):def __init__(self, outer_nc, inner_nc, in_planes=None, dropout=False,submodule=None, outermost=False, innermost=False, alpha=0.2, norm_mode='batch'):super(UNetSkipConnectionBlock, self).__init__()down_norm = nn.BatchNorm2d(inner_nc)up_norm = nn.BatchNorm2d(outer_nc)use_bias = Falseif norm_mode == 'instance':down_norm = nn.BatchNorm2d(inner_nc, affine=False)up_norm = nn.BatchNorm2d(outer_nc, affine=False)use_bias = Trueif in_planes is None:in_planes = outer_ncdown_conv = nn.Conv2d(in_planes, inner_nc, kernel_size=4,stride=2, padding=1, has_bias=use_bias, pad_mode='pad')down_relu = nn.LeakyReLU(alpha)up_relu = nn.ReLU()if outermost:up_conv = nn.Conv2dTranspose(inner_nc * 2, outer_nc,kernel_size=4, stride=2,padding=1, pad_mode='pad')down = [down_conv]up = [up_relu, up_conv, nn.Tanh()]model = down + [submodule] + upelif innermost:up_conv = nn.Conv2dTranspose(inner_nc, outer_nc,kernel_size=4, stride=2,padding=1, has_bias=use_bias, pad_mode='pad')down = [down_relu, down_conv]up = [up_relu, up_conv, up_norm]model = down + upelse:up_conv = nn.Conv2dTranspose(inner_nc * 2, outer_nc,kernel_size=4, stride=2,padding=1, has_bias=use_bias, pad_mode='pad')down = [down_relu, down_conv, down_norm]up = [up_relu, up_conv, up_norm]model = down + [submodule] + upif dropout:model.append(nn.Dropout(p=0.5))self.model = nn.SequentialCell(model)self.skip_connections = not outermostdef construct(self, x):out = self.model(x)if self.skip_connections:out = ops.concat((out, x), axis=1)return out
基于UNet的生成器
class UNetGenerator(nn.Cell):def __init__(self, in_planes, out_planes, ngf=64, n_layers=8, norm_mode='bn', dropout=False):super(UNetGenerator, self).__init__()unet_block = UNetSkipConnectionBlock(ngf * 8, ngf * 8, in_planes=None, submodule=None,norm_mode=norm_mode, innermost=True)for _ in range(n_layers - 5):unet_block = UNetSkipConnectionBlock(ngf * 8, ngf * 8, in_planes=None, submodule=unet_block,norm_mode=norm_mode, dropout=dropout)unet_block = UNetSkipConnectionBlock(ngf * 4, ngf * 8, in_planes=None, submodule=unet_block,norm_mode=norm_mode)unet_block = UNetSkipConnectionBlock(ngf * 2, ngf * 4, in_planes=None, submodule=unet_block,norm_mode=norm_mode)unet_block = UNetSkipConnectionBlock(ngf, ngf * 2, in_planes=None, submodule=unet_block,norm_mode=norm_mode)self.model = UNetSkipConnectionBlock(out_planes, ngf, in_planes=in_planes, submodule=unet_block,outermost=True, norm_mode=norm_mode)def construct(self, x):return self.model(x)

原始cGAN的输入是条件x和噪声z两种信息,这里的生成器只使用了条件信息,因此不能生成多样性的结果。因此Pix2Pix在训练和测试时都使用了dropout,这样可以生成多样性的结果。

基于PatchGAN的判别器

判别器使用的PatchGAN结构,可看做卷积。生成的矩阵中的每个点代表原图的一小块区域(patch)。通过矩阵中的各个值来判断原图中对应每个Patch的真假。

import mindspore.nn as nnclass ConvNormRelu(nn.Cell):def __init__(self,in_planes,out_planes,kernel_size=4,stride=2,alpha=0.2,norm_mode='batch',pad_mode='CONSTANT',use_relu=True,padding=None):super(ConvNormRelu, self).__init__()norm = nn.BatchNorm2d(out_planes)if norm_mode == 'instance':norm = nn.BatchNorm2d(out_planes, affine=False)has_bias = (norm_mode == 'instance')if not padding:padding = (kernel_size - 1) // 2if pad_mode == 'CONSTANT':conv = nn.Conv2d(in_planes, out_planes, kernel_size, stride, pad_mode='pad',has_bias=has_bias, padding=padding)layers = [conv, norm]else:paddings = ((0, 0), (0, 0), (padding, padding), (padding, padding))pad = nn.Pad(paddings=paddings, mode=pad_mode)conv = nn.Conv2d(in_planes, out_planes, kernel_size, stride, pad_mode='pad', has_bias=has_bias)layers = [pad, conv, norm]if use_relu:relu = nn.ReLU()if alpha > 0:relu = nn.LeakyReLU(alpha)layers.append(relu)self.features = nn.SequentialCell(layers)def construct(self, x):output = self.features(x)return outputclass Discriminator(nn.Cell):def __init__(self, in_planes=3, ndf=64, n_layers=3, alpha=0.2, norm_mode='batch'):super(Discriminator, self).__init__()kernel_size = 4layers = [nn.Conv2d(in_planes, ndf, kernel_size, 2, pad_mode='pad', padding=1),nn.LeakyReLU(alpha)]nf_mult = ndffor i in range(1, n_layers):nf_mult_prev = nf_multnf_mult = min(2 ** i, 8) * ndflayers.append(ConvNormRelu(nf_mult_prev, nf_mult, kernel_size, 2, alpha, norm_mode, padding=1))nf_mult_prev = nf_multnf_mult = min(2 ** n_layers, 8) * ndflayers.append(ConvNormRelu(nf_mult_prev, nf_mult, kernel_size, 1, alpha, norm_mode, padding=1))layers.append(nn.Conv2d(nf_mult, 1, kernel_size, 1, pad_mode='pad', padding=1))self.features = nn.SequentialCell(layers)def construct(self, x, y):x_y = ops.concat((x, y), axis=1)output = self.features(x_y)return output

Pix2Pix的生成器和判别器初始化

实例化Pix2Pix生成器和判别器。

import mindspore.nn as nn
from mindspore.common import initializer as initg_in_planes = 3
g_out_planes = 3
g_ngf = 64
g_layers = 8
d_in_planes = 6
d_ndf = 64
d_layers = 3
alpha = 0.2
init_gain = 0.02
init_type = 'normal'net_generator = UNetGenerator(in_planes=g_in_planes, out_planes=g_out_planes,ngf=g_ngf, n_layers=g_layers)
for _, cell in net_generator.cells_and_names():if isinstance(cell, (nn.Conv2d, nn.Conv2dTranspose)):if init_type == 'normal':cell.weight.set_data(init.initializer(init.Normal(init_gain), cell.weight.shape))elif init_type == 'xavier':cell.weight.set_data(init.initializer(init.XavierUniform(init_gain), cell.weight.shape))elif init_type == 'constant':cell.weight.set_data(init.initializer(0.001, cell.weight.shape))else:raise NotImplementedError('initialization method [%s] is not implemented' % init_type)elif isinstance(cell, nn.BatchNorm2d):cell.gamma.set_data(init.initializer('ones', cell.gamma.shape))cell.beta.set_data(init.initializer('zeros', cell.beta.shape))net_discriminator = Discriminator(in_planes=d_in_planes, ndf=d_ndf,alpha=alpha, n_layers=d_layers)
for _, cell in net_discriminator.cells_and_names():if isinstance(cell, (nn.Conv2d, nn.Conv2dTranspose)):if init_type == 'normal':cell.weight.set_data(init.initializer(init.Normal(init_gain), cell.weight.shape))elif init_type == 'xavier':cell.weight.set_data(init.initializer(init.XavierUniform(init_gain), cell.weight.shape))elif init_type == 'constant':cell.weight.set_data(init.initializer(0.001, cell.weight.shape))else:raise NotImplementedError('initialization method [%s] is not implemented' % init_type)elif isinstance(cell, nn.BatchNorm2d):cell.gamma.set_data(init.initializer('ones', cell.gamma.shape))cell.beta.set_data(init.initializer('zeros', cell.beta.shape))class Pix2Pix(nn.Cell):"""Pix2Pix模型网络"""def __init__(self, discriminator, generator):super(Pix2Pix, self).__init__(auto_prefix=True)self.net_discriminator = discriminatorself.net_generator = generatordef construct(self, reala):fakeb = self.net_generator(reala)return fakeb

训练

训练分为两个主要部分:训练判别器和训练生成器。训练判别器的目的是最大程度地提高判别图像真伪的概率。训练生成器是希望能产生更好的虚假图像。在这两个部分中,分别获取训练过程中的损失,并在每个周期结束时进行统计。

下面进行训练:

%%time
import numpy as np
import os
import datetime
from mindspore import value_and_grad, Tensorepoch_num = 100
ckpt_dir = "results/ckpt"
dataset_size = 400
val_pic_size = 256
lr = 0.0002
n_epochs = 100
n_epochs_decay = 100def get_lr():lrs = [lr] * dataset_size * n_epochslr_epoch = 0for epoch in range(n_epochs_decay):lr_epoch = lr * (n_epochs_decay - epoch) / n_epochs_decaylrs += [lr_epoch] * dataset_sizelrs += [lr_epoch] * dataset_size * (epoch_num - n_epochs_decay - n_epochs)return Tensor(np.array(lrs).astype(np.float32))dataset = ds.MindDataset("./dataset/dataset_pix2pix/train.mindrecord", columns_list=["input_images", "target_images"], shuffle=True, num_parallel_workers=1)
steps_per_epoch = dataset.get_dataset_size()
loss_f = nn.BCEWithLogitsLoss()
l1_loss = nn.L1Loss()def forword_dis(reala, realb):lambda_dis = 0.5fakeb = net_generator(reala)pred0 = net_discriminator(reala, fakeb)pred1 = net_discriminator(reala, realb)loss_d = loss_f(pred1, ops.ones_like(pred1)) + loss_f(pred0, ops.zeros_like(pred0))loss_dis = loss_d * lambda_disreturn loss_disdef forword_gan(reala, realb):lambda_gan = 0.5lambda_l1 = 100fakeb = net_generator(reala)pred0 = net_discriminator(reala, fakeb)loss_1 = loss_f(pred0, ops.ones_like(pred0))loss_2 = l1_loss(fakeb, realb)loss_gan = loss_1 * lambda_gan + loss_2 * lambda_l1return loss_gand_opt = nn.Adam(net_discriminator.trainable_params(), learning_rate=get_lr(),beta1=0.5, beta2=0.999, loss_scale=1)
g_opt = nn.Adam(net_generator.trainable_params(), learning_rate=get_lr(),beta1=0.5, beta2=0.999, loss_scale=1)grad_d = value_and_grad(forword_dis, None, net_discriminator.trainable_params())
grad_g = value_and_grad(forword_gan, None, net_generator.trainable_params())def train_step(reala, realb):loss_dis, d_grads = grad_d(reala, realb)loss_gan, g_grads = grad_g(reala, realb)d_opt(d_grads)g_opt(g_grads)return loss_dis, loss_ganif not os.path.isdir(ckpt_dir):os.makedirs(ckpt_dir)g_losses = []
d_losses = []
data_loader = dataset.create_dict_iterator(output_numpy=True, num_epochs=epoch_num)for epoch in range(epoch_num):for i, data in enumerate(data_loader):start_time = datetime.datetime.now()input_image = Tensor(data["input_images"])target_image = Tensor(data["target_images"])dis_loss, gen_loss = train_step(input_image, target_image)end_time = datetime.datetime.now()delta = (end_time - start_time).microsecondsif i % 2 == 0:print("ms per step:{:.2f}  epoch:{}/{}  step:{}/{}  Dloss:{:.4f}  Gloss:{:.4f} ".format((delta / 1000), (epoch + 1), (epoch_num), i, steps_per_epoch, float(dis_loss), float(gen_loss)))d_losses.append(dis_loss.asnumpy())g_losses.append(gen_loss.asnumpy())if (epoch + 1) == epoch_num:mindspore.save_checkpoint(net_generator, ckpt_dir + "Generator.ckpt")
ms per step:204.75  epoch:1/100  step:0/25  Dloss:0.6897  Gloss:37.1031 
ms per step:110.92  epoch:1/100  step:2/25  Dloss:0.6353  Gloss:32.0685 
ms per step:109.04  epoch:1/100  step:4/25  Dloss:0.4980  Gloss:38.4479 
ms per step:109.70  epoch:1/100  step:6/25  Dloss:0.4706  Gloss:40.6548 
ms per step:111.17  epoch:1/100  step:8/25  Dloss:0.3736  Gloss:37.2207 
ms per step:116.04  epoch:1/100  step:10/25  Dloss:0.3313  Gloss:39.9487 
ms per step:114.72  epoch:1/100  step:12/25  Dloss:0.5754  Gloss:37.0347 
ms per step:115.50  epoch:1/100  step:14/25  Dloss:0.2367  Gloss:32.5301 
ms per step:115.18  epoch:1/100  step:16/25  Dloss:0.2420  Gloss:36.9372 
ms per step:116.61  epoch:1/100  step:18/25  Dloss:0.3123  Gloss:39.1208 
ms per step:116.03  epoch:1/100  step:20/25  Dloss:0.3024  Gloss:32.9262 
ms per step:114.12  epoch:1/100  step:22/25  Dloss:0.2165  Gloss:38.9377 
ms per step:115.46  epoch:1/100  step:24/25  Dloss:0.1766  Gloss:36.7368 
ms per step:110.67  epoch:2/100  step:0/25  Dloss:0.3093  Gloss:36.5221 
ms per step:106.14  epoch:2/100  step:2/25  Dloss:0.1656  Gloss:35.4346 
ms per step:111.77  epoch:2/100  step:4/25  Dloss:0.1827  Gloss:35.6615 
ms per step:112.56  epoch:2/100  step:6/25  Dloss:0.8171  Gloss:35.8147 
ms per step:109.08  epoch:2/100  step:8/25  Dloss:0.3058  Gloss:33.9187 
ms per step:112.14  epoch:2/100  step:10/25  Dloss:0.8390  Gloss:42.6125 
ms per step:156.64  epoch:2/100  step:12/25  Dloss:0.3088  Gloss:31.6007 
ms per step:138.70  epoch:2/100  step:14/25  Dloss:0.2673  Gloss:37.3562 
ms per step:165.71  epoch:2/100  step:16/25  Dloss:0.2179  Gloss:35.6225 
ms per step:161.63  epoch:2/100  step:18/25  Dloss:0.3507  Gloss:37.7050 
ms per step:124.58  epoch:2/100  step:20/25  Dloss:0.2014  Gloss:37.8277 
ms per step:109.08  epoch:2/100  step:22/25  Dloss:4.5342  Gloss:39.6272 
ms per step:106.03  epoch:2/100  step:24/25  Dloss:0.2985  Gloss:39.3276 
ms per step:112.50  epoch:3/100  step:0/25  Dloss:0.3666  Gloss:39.4434 
ms per step:127.56  epoch:3/100  step:2/25  Dloss:0.3360  Gloss:37.8373 
ms per step:117.78  epoch:3/100  step:4/25  Dloss:0.3265  Gloss:35.4772 
ms per step:116.28  epoch:3/100  step:6/25  Dloss:0.2673  Gloss:38.2814 
ms per step:117.82  epoch:3/100  step:8/25  Dloss:0.2192  Gloss:36.0040 
ms per step:127.89  epoch:3/100  step:10/25  Dloss:0.5457  Gloss:35.2886 
ms per step:115.94  epoch:3/100  step:12/25  Dloss:0.3134  Gloss:33.9577 
ms per step:113.05  epoch:3/100  step:14/25  Dloss:0.2128  Gloss:36.5794 
ms per step:113.79  epoch:3/100  step:16/25  Dloss:0.1982  Gloss:38.8131 
ms per step:145.94  epoch:3/100  step:18/25  Dloss:0.3664  Gloss:33.1946 
ms per step:123.02  epoch:3/100  step:20/25  Dloss:0.3942  Gloss:35.1858 
ms per step:111.32  epoch:3/100  step:22/25  Dloss:0.2949  Gloss:33.3575 
ms per step:107.27  epoch:3/100  step:24/25  Dloss:0.3199  Gloss:35.6960 
ms per step:114.69  epoch:4/100  step:0/25  Dloss:0.3226  Gloss:30.9485 
ms per step:106.25  epoch:4/100  step:2/25  Dloss:0.6456  Gloss:34.4001 
ms per step:104.81  epoch:4/100  step:4/25  Dloss:0.3166  Gloss:33.7080 
ms per step:99.45  epoch:4/100  step:6/25  Dloss:0.2429  Gloss:36.3314 
ms per step:101.94  epoch:4/100  step:8/25  Dloss:0.3957  Gloss:36.2481 
ms per step:106.84  epoch:4/100  step:10/25  Dloss:0.3102  Gloss:34.8957 
ms per step:117.80  epoch:4/100  step:12/25  Dloss:0.2448  Gloss:34.8868 
ms per step:118.18  epoch:4/100  step:14/25  Dloss:0.4172  Gloss:35.7913 
ms per step:114.14  epoch:4/100  step:16/25  Dloss:0.7911  Gloss:32.5833 
ms per step:113.04  epoch:4/100  step:18/25  Dloss:0.2795  Gloss:30.4272 
ms per step:108.48  epoch:4/100  step:20/25  Dloss:0.2095  Gloss:34.1284 
ms per step:112.32  epoch:4/100  step:22/25  Dloss:0.2670  Gloss:34.6068 
ms per step:101.79  epoch:4/100  step:24/25  Dloss:0.5286  Gloss:34.2540 
ms per step:102.98  epoch:5/100  step:0/25  Dloss:0.4048  Gloss:39.6592 
ms per step:107.02  epoch:5/100  step:2/25  Dloss:0.3032  Gloss:33.6503 
ms per step:112.57  epoch:5/100  step:4/25  Dloss:0.1963  Gloss:36.0851 
ms per step:101.16  epoch:5/100  step:6/25  Dloss:0.2779  Gloss:35.6111 
ms per step:109.16  epoch:5/100  step:8/25  Dloss:0.1857  Gloss:34.5232 
ms per step:110.14  epoch:5/100  step:10/25  Dloss:0.5421  Gloss:30.3469 
ms per step:110.88  epoch:5/100  step:12/25  Dloss:0.3997  Gloss:30.9148 
ms per step:108.31  epoch:5/100  step:14/25  Dloss:0.2931  Gloss:33.7271 
ms per step:112.18  epoch:5/100  step:16/25  Dloss:0.2698  Gloss:30.7518 
ms per step:112.77  epoch:5/100  step:18/25  Dloss:0.2204  Gloss:35.0685 
ms per step:107.06  epoch:5/100  step:20/25  Dloss:0.2335  Gloss:35.6151 
ms per step:100.25  epoch:5/100  step:22/25  Dloss:0.2479  Gloss:35.1270 
ms per step:105.97  epoch:5/100  step:24/25  Dloss:0.2322  Gloss:33.7123 
ms per step:99.31  epoch:6/100  step:0/25  Dloss:0.2245  Gloss:30.2303 
ms per step:107.41  epoch:6/100  step:2/25  Dloss:0.1197  Gloss:34.8541 
ms per step:102.88  epoch:6/100  step:4/25  Dloss:0.8449  Gloss:37.6315 
ms per step:111.83  epoch:6/100  step:6/25  Dloss:0.2133  Gloss:33.7175 
ms per step:103.20  epoch:6/100  step:8/25  Dloss:0.2666  Gloss:30.4968 
ms per step:108.35  epoch:6/100  step:10/25  Dloss:0.1703  Gloss:34.2822 
ms per step:117.95  epoch:6/100  step:12/25  Dloss:0.1685  Gloss:33.4703 
ms per step:111.74  epoch:6/100  step:14/25  Dloss:0.1708  Gloss:35.2430 
ms per step:118.85  epoch:6/100  step:16/25  Dloss:0.1398  Gloss:35.2619 
ms per step:108.81  epoch:6/100  step:18/25  Dloss:0.1568  Gloss:33.8083 
ms per step:115.50  epoch:6/100  step:20/25  Dloss:0.3280  Gloss:36.0523 
ms per step:108.38  epoch:6/100  step:22/25  Dloss:0.2311  Gloss:36.7951 
ms per step:100.83  epoch:6/100  step:24/25  Dloss:0.1700  Gloss:33.5755 
ms per step:107.63  epoch:7/100  step:0/25  Dloss:0.2986  Gloss:34.4059 
ms per step:102.86  epoch:7/100  step:2/25  Dloss:0.1548  Gloss:37.1741 
ms per step:105.17  epoch:7/100  step:4/25  Dloss:0.2311  Gloss:35.0723 
ms per step:102.77  epoch:7/100  step:6/25  Dloss:0.4507  Gloss:36.0862 
ms per step:105.28  epoch:7/100  step:8/25  Dloss:0.4507  Gloss:30.9377 
ms per step:118.02  epoch:7/100  step:10/25  Dloss:0.1870  Gloss:34.9944 
ms per step:123.09  epoch:7/100  step:12/25  Dloss:0.1602  Gloss:30.1066 
ms per step:116.36  epoch:7/100  step:14/25  Dloss:0.1795  Gloss:36.0846 
ms per step:107.29  epoch:7/100  step:16/25  Dloss:0.1186  Gloss:35.2313 
ms per step:114.19  epoch:7/100  step:18/25  Dloss:0.1104  Gloss:33.9729 
ms per step:111.15  epoch:7/100  step:20/25  Dloss:0.3128  Gloss:37.1534 
ms per step:100.68  epoch:7/100  step:22/25  Dloss:0.2224  Gloss:34.6298 
ms per step:101.16  epoch:7/100  step:24/25  Dloss:0.0838  Gloss:35.4592 
ms per step:101.17  epoch:8/100  step:0/25  Dloss:0.2538  Gloss:35.1292 
ms per step:105.37  epoch:8/100  step:2/25  Dloss:0.1362  Gloss:36.2886 
ms per step:105.59  epoch:8/100  step:4/25  Dloss:0.1124  Gloss:36.5091 
ms per step:104.59  epoch:8/100  step:6/25  Dloss:0.1557  Gloss:32.0394 
ms per step:100.29  epoch:8/100  step:8/25  Dloss:0.1529  Gloss:36.9445 
ms per step:107.35  epoch:8/100  step:10/25  Dloss:0.1066  Gloss:36.0111 
ms per step:116.61  epoch:8/100  step:12/25  Dloss:0.1983  Gloss:31.5163 
ms per step:119.83  epoch:8/100  step:14/25  Dloss:0.1146  Gloss:35.5416 
ms per step:116.71  epoch:8/100  step:16/25  Dloss:0.2489  Gloss:33.4765 
ms per step:107.50  epoch:8/100  step:18/25  Dloss:1.1940  Gloss:37.1177 
ms per step:99.72  epoch:8/100  step:20/25  Dloss:0.2300  Gloss:34.8493 
ms per step:99.67  epoch:8/100  step:22/25  Dloss:0.1331  Gloss:34.4320 
ms per step:103.94  epoch:8/100  step:24/25  Dloss:0.1599  Gloss:33.5074 
ms per step:105.40  epoch:9/100  step:0/25  Dloss:0.2450  Gloss:34.0505 
ms per step:108.44  epoch:9/100  step:2/25  Dloss:0.1436  Gloss:35.5095 
ms per step:105.19  epoch:9/100  step:4/25  Dloss:0.1030  Gloss:34.1643 
ms per step:99.91  epoch:9/100  step:6/25  Dloss:0.1102  Gloss:29.2541 
ms per step:108.14  epoch:9/100  step:8/25  Dloss:0.1679  Gloss:34.8311 
ms per step:108.99  epoch:9/100  step:10/25  Dloss:0.1670  Gloss:33.6990 
ms per step:113.90  epoch:9/100  step:12/25  Dloss:0.0852  Gloss:35.4889 
ms per step:113.92  epoch:9/100  step:14/25  Dloss:0.1765  Gloss:38.7903 
ms per step:108.45  epoch:9/100  step:16/25  Dloss:0.1475  Gloss:32.4022 
ms per step:106.38  epoch:9/100  step:18/25  Dloss:0.1035  Gloss:35.0014 
ms per step:106.81  epoch:9/100  step:20/25  Dloss:0.2029  Gloss:31.7662 
ms per step:97.29  epoch:9/100  step:22/25  Dloss:0.2387  Gloss:36.9439 
ms per step:99.65  epoch:9/100  step:24/25  Dloss:0.6291  Gloss:33.8078 
ms per step:96.38  epoch:10/100  step:0/25  Dloss:0.7573  Gloss:37.2553 
ms per step:135.43  epoch:10/100  step:2/25  Dloss:0.1455  Gloss:37.1506 
ms per step:97.36  epoch:10/100  step:4/25  Dloss:0.2290  Gloss:32.2825 
ms per step:97.25  epoch:10/100  step:6/25  Dloss:0.2324  Gloss:36.1145 
ms per step:96.77  epoch:10/100  step:8/25  Dloss:0.1912  Gloss:34.6552 
ms per step:103.31  epoch:10/100  step:10/25  Dloss:0.1369  Gloss:35.1498 
ms per step:137.80  epoch:10/100  step:12/25  Dloss:0.1725  Gloss:33.3661 
ms per step:111.68  epoch:10/100  step:14/25  Dloss:0.2208  Gloss:34.4513 
ms per step:141.62  epoch:10/100  step:16/25  Dloss:0.2574  Gloss:33.2548 
ms per step:149.45  epoch:10/100  step:18/25  Dloss:0.2365  Gloss:28.9793 
ms per step:128.19  epoch:10/100  step:20/25  Dloss:0.2035  Gloss:32.4743 
ms per step:125.08  epoch:10/100  step:22/25  Dloss:0.7574  Gloss:36.6609 
ms per step:99.11  epoch:10/100  step:24/25  Dloss:0.1860  Gloss:34.6567 
ms per step:123.55  epoch:11/100  step:0/25  Dloss:0.2223  Gloss:28.5527 
ms per step:95.78  epoch:11/100  step:2/25  Dloss:0.1632  Gloss:33.4401 
ms per step:149.95  epoch:11/100  step:4/25  Dloss:0.2714  Gloss:33.2374 
ms per step:149.31  epoch:11/100  step:6/25  Dloss:0.1149  Gloss:32.7205 
ms per step:122.46  epoch:11/100  step:8/25  Dloss:0.1357  Gloss:32.8428 
ms per step:101.08  epoch:11/100  step:10/25  Dloss:0.1149  Gloss:30.7533 
ms per step:101.09  epoch:11/100  step:12/25  Dloss:0.1705  Gloss:32.3486 
ms per step:129.94  epoch:11/100  step:14/25  Dloss:0.1711  Gloss:30.2202 
ms per step:103.29  epoch:11/100  step:16/25  Dloss:0.2044  Gloss:33.3232 
ms per step:95.99  epoch:11/100  step:18/25  Dloss:0.5314  Gloss:36.5913 
ms per step:95.58  epoch:11/100  step:20/25  Dloss:0.3198  Gloss:33.0979 
ms per step:95.93  epoch:11/100  step:22/25  Dloss:0.2615  Gloss:34.2509 
ms per step:95.08  epoch:11/100  step:24/25  Dloss:0.1590  Gloss:36.7118 
ms per step:95.77  epoch:12/100  step:0/25  Dloss:0.1776  Gloss:32.5703 
ms per step:95.53  epoch:12/100  step:2/25  Dloss:0.2224  Gloss:31.7528 
ms per step:95.47  epoch:12/100  step:4/25  Dloss:0.1915  Gloss:32.9494 
ms per step:95.50  epoch:12/100  step:6/25  Dloss:0.1451  Gloss:35.0644 
ms per step:95.43  epoch:12/100  step:8/25  Dloss:0.1969  Gloss:35.8737 
ms per step:105.98  epoch:12/100  step:10/25  Dloss:0.3251  Gloss:31.5041 
ms per step:106.46  epoch:12/100  step:12/25  Dloss:0.2659  Gloss:31.5948 
ms per step:105.40  epoch:12/100  step:14/25  Dloss:0.1684  Gloss:30.4842 
ms per step:102.45  epoch:12/100  step:16/25  Dloss:0.2178  Gloss:32.0689 
ms per step:95.56  epoch:12/100  step:18/25  Dloss:0.1996  Gloss:36.5052 
ms per step:95.78  epoch:12/100  step:20/25  Dloss:0.1974  Gloss:34.0614 
ms per step:106.77  epoch:12/100  step:22/25  Dloss:0.1365  Gloss:28.7357 
ms per step:95.95  epoch:12/100  step:24/25  Dloss:0.1066  Gloss:32.5434 
ms per step:96.25  epoch:13/100  step:0/25  Dloss:0.1326  Gloss:34.3436 
ms per step:95.90  epoch:13/100  step:2/25  Dloss:0.2720  Gloss:35.8570 
ms per step:95.71  epoch:13/100  step:4/25  Dloss:0.4157  Gloss:32.6792 
ms per step:96.05  epoch:13/100  step:6/25  Dloss:0.4953  Gloss:35.1763 
ms per step:96.15  epoch:13/100  step:8/25  Dloss:0.1667  Gloss:28.0880 
ms per step:100.28  epoch:13/100  step:10/25  Dloss:0.1568  Gloss:32.0826 
ms per step:100.30  epoch:13/100  step:12/25  Dloss:0.1554  Gloss:31.7710 
ms per step:101.40  epoch:13/100  step:14/25  Dloss:0.1292  Gloss:31.0753 
ms per step:100.00  epoch:13/100  step:16/25  Dloss:0.1329  Gloss:32.0049 
ms per step:96.21  epoch:13/100  step:18/25  Dloss:0.0770  Gloss:30.7780 
ms per step:95.55  epoch:13/100  step:20/25  Dloss:0.1226  Gloss:33.3695 
ms per step:95.54  epoch:13/100  step:22/25  Dloss:0.3987  Gloss:32.6538 
ms per step:95.69  epoch:13/100  step:24/25  Dloss:0.2782  Gloss:31.2371 
ms per step:97.27  epoch:14/100  step:0/25  Dloss:0.3108  Gloss:31.3417 
ms per step:96.36  epoch:14/100  step:2/25  Dloss:0.2313  Gloss:30.3941 
ms per step:95.96  epoch:14/100  step:4/25  Dloss:0.1593  Gloss:30.4305 
ms per step:96.17  epoch:14/100  step:6/25  Dloss:0.1321  Gloss:29.9868 
ms per step:95.92  epoch:14/100  step:8/25  Dloss:0.1414  Gloss:29.1535 
ms per step:99.88  epoch:14/100  step:10/25  Dloss:0.1748  Gloss:27.6295 
ms per step:102.80  epoch:14/100  step:12/25  Dloss:0.0851  Gloss:30.4776 
ms per step:104.72  epoch:14/100  step:14/25  Dloss:0.1030  Gloss:33.5051 
ms per step:105.77  epoch:14/100  step:16/25  Dloss:0.2157  Gloss:28.4780 
ms per step:99.08  epoch:14/100  step:18/25  Dloss:0.3448  Gloss:31.6692 
ms per step:96.08  epoch:14/100  step:20/25  Dloss:0.6413  Gloss:33.1093 
ms per step:94.78  epoch:14/100  step:22/25  Dloss:0.3616  Gloss:31.4116 
ms per step:96.94  epoch:14/100  step:24/25  Dloss:0.2936  Gloss:32.5213 
ms per step:96.04  epoch:15/100  step:0/25  Dloss:0.2943  Gloss:29.0930 
ms per step:101.20  epoch:15/100  step:2/25  Dloss:0.2802  Gloss:27.3098 
ms per step:97.18  epoch:15/100  step:4/25  Dloss:0.2515  Gloss:29.4319 
ms per step:97.17  epoch:15/100  step:6/25  Dloss:0.1181  Gloss:32.0332 
ms per step:98.63  epoch:15/100  step:8/25  Dloss:0.1632  Gloss:29.9418 
ms per step:103.15  epoch:15/100  step:10/25  Dloss:0.2113  Gloss:32.2050 
ms per step:101.50  epoch:15/100  step:12/25  Dloss:0.4008  Gloss:27.3341 
ms per step:106.53  epoch:15/100  step:14/25  Dloss:0.3969  Gloss:29.2508 
ms per step:137.32  epoch:15/100  step:16/25  Dloss:0.2252  Gloss:31.2349 
ms per step:109.99  epoch:15/100  step:18/25  Dloss:0.2692  Gloss:26.9864 
ms per step:97.49  epoch:15/100  step:20/25  Dloss:0.1570  Gloss:28.7567 
ms per step:98.15  epoch:15/100  step:22/25  Dloss:0.2171  Gloss:27.2928 
ms per step:102.67  epoch:15/100  step:24/25  Dloss:0.1300  Gloss:32.1528 
ms per step:99.73  epoch:16/100  step:0/25  Dloss:0.1215  Gloss:28.0065 
ms per step:98.55  epoch:16/100  step:2/25  Dloss:0.3855  Gloss:31.0583 
ms per step:97.59  epoch:16/100  step:4/25  Dloss:0.3456  Gloss:29.2974 
ms per step:96.91  epoch:16/100  step:6/25  Dloss:0.6300  Gloss:29.3516 
ms per step:98.30  epoch:16/100  step:8/25  Dloss:0.2407  Gloss:26.1584 
ms per step:131.25  epoch:16/100  step:10/25  Dloss:0.1509  Gloss:30.9893 
ms per step:106.43  epoch:16/100  step:12/25  Dloss:0.1725  Gloss:31.1133 
ms per step:107.43  epoch:16/100  step:14/25  Dloss:0.1371  Gloss:28.4015 
ms per step:107.44  epoch:16/100  step:16/25  Dloss:0.5244  Gloss:27.1161 
ms per step:99.51  epoch:16/100  step:18/25  Dloss:0.6425  Gloss:29.8100 
ms per step:100.53  epoch:16/100  step:20/25  Dloss:0.3396  Gloss:30.0259 
ms per step:105.17  epoch:16/100  step:22/25  Dloss:0.5974  Gloss:30.8600 
ms per step:100.24  epoch:16/100  step:24/25  Dloss:0.2338  Gloss:23.1849 
ms per step:99.72  epoch:17/100  step:0/25  Dloss:0.1453  Gloss:28.1197 
ms per step:96.99  epoch:17/100  step:2/25  Dloss:0.1593  Gloss:27.4453 
ms per step:98.94  epoch:17/100  step:4/25  Dloss:0.0717  Gloss:30.1956 
ms per step:99.35  epoch:17/100  step:6/25  Dloss:0.1333  Gloss:29.3688 
ms per step:97.92  epoch:17/100  step:8/25  Dloss:0.2118  Gloss:29.1178 
ms per step:109.60  epoch:17/100  step:10/25  Dloss:0.1913  Gloss:24.0773 
ms per step:107.75  epoch:17/100  step:12/25  Dloss:0.2053  Gloss:26.8610 
ms per step:110.58  epoch:17/100  step:14/25  Dloss:0.4358  Gloss:25.2621 
ms per step:105.61  epoch:17/100  step:16/25  Dloss:0.2057  Gloss:26.8026 
ms per step:97.85  epoch:17/100  step:18/25  Dloss:0.1685  Gloss:28.7992 
ms per step:97.79  epoch:17/100  step:20/25  Dloss:0.2067  Gloss:28.2258 
ms per step:97.57  epoch:17/100  step:22/25  Dloss:0.3917  Gloss:32.2305 
ms per step:98.86  epoch:17/100  step:24/25  Dloss:0.4569  Gloss:27.4586 
ms per step:103.15  epoch:18/100  step:0/25  Dloss:0.3312  Gloss:31.0248 
ms per step:98.32  epoch:18/100  step:2/25  Dloss:0.2100  Gloss:27.5618 
ms per step:101.75  epoch:18/100  step:4/25  Dloss:0.1853  Gloss:27.0157 
ms per step:102.89  epoch:18/100  step:6/25  Dloss:0.1962  Gloss:24.5970 
ms per step:102.29  epoch:18/100  step:8/25  Dloss:0.1631  Gloss:30.0126 
ms per step:106.48  epoch:18/100  step:10/25  Dloss:0.2848  Gloss:28.3787 
ms per step:117.61  epoch:18/100  step:12/25  Dloss:0.4688  Gloss:24.4731 
ms per step:112.87  epoch:18/100  step:14/25  Dloss:0.3016  Gloss:25.3336 
ms per step:107.60  epoch:18/100  step:16/25  Dloss:0.1850  Gloss:27.0883 
ms per step:101.59  epoch:18/100  step:18/25  Dloss:0.1652  Gloss:26.5043 
ms per step:98.58  epoch:18/100  step:20/25  Dloss:0.1916  Gloss:29.1534 
ms per step:101.63  epoch:18/100  step:22/25  Dloss:0.1935  Gloss:28.6713 
ms per step:97.22  epoch:18/100  step:24/25  Dloss:0.2222  Gloss:29.5466 
ms per step:98.96  epoch:19/100  step:0/25  Dloss:0.4617  Gloss:26.9988 
ms per step:100.26  epoch:19/100  step:2/25  Dloss:0.2117  Gloss:25.0168 
ms per step:102.04  epoch:19/100  step:4/25  Dloss:0.2702  Gloss:23.6539 
ms per step:102.21  epoch:19/100  step:6/25  Dloss:0.2227  Gloss:27.0165 
ms per step:99.51  epoch:19/100  step:8/25  Dloss:0.2021  Gloss:27.7990 
ms per step:133.01  epoch:19/100  step:10/25  Dloss:0.4087  Gloss:25.5694 
ms per step:136.31  epoch:19/100  step:12/25  Dloss:0.3771  Gloss:26.0101 
ms per step:119.35  epoch:19/100  step:14/25  Dloss:0.3873  Gloss:26.1754 
ms per step:107.84  epoch:19/100  step:16/25  Dloss:0.1979  Gloss:28.9217 
ms per step:104.08  epoch:19/100  step:18/25  Dloss:0.1825  Gloss:27.0119 
ms per step:110.75  epoch:19/100  step:20/25  Dloss:0.1285  Gloss:27.8270 
ms per step:100.82  epoch:19/100  step:22/25  Dloss:0.1690  Gloss:26.4484 
ms per step:101.63  epoch:19/100  step:24/25  Dloss:0.1743  Gloss:25.0522 
ms per step:101.40  epoch:20/100  step:0/25  Dloss:0.3100  Gloss:25.8043 
ms per step:104.71  epoch:20/100  step:2/25  Dloss:0.1535  Gloss:24.0780 
ms per step:106.20  epoch:20/100  step:4/25  Dloss:0.2306  Gloss:25.1991 
ms per step:106.81  epoch:20/100  step:6/25  Dloss:0.3823  Gloss:24.3071 
ms per step:100.66  epoch:20/100  step:8/25  Dloss:0.3268  Gloss:25.5527 
ms per step:110.43  epoch:20/100  step:10/25  Dloss:0.2606  Gloss:26.6634 
ms per step:110.04  epoch:20/100  step:12/25  Dloss:0.3721  Gloss:24.7309 
ms per step:114.23  epoch:20/100  step:14/25  Dloss:0.2918  Gloss:25.2503 
ms per step:107.06  epoch:20/100  step:16/25  Dloss:0.5828  Gloss:26.4883 
ms per step:100.38  epoch:20/100  step:18/25  Dloss:0.3215  Gloss:26.1511 
ms per step:107.65  epoch:20/100  step:20/25  Dloss:0.3962  Gloss:26.0596 
ms per step:105.98  epoch:20/100  step:22/25  Dloss:0.2283  Gloss:24.3636 
ms per step:112.09  epoch:20/100  step:24/25  Dloss:0.1482  Gloss:23.4873 
ms per step:107.53  epoch:21/100  step:0/25  Dloss:0.1888  Gloss:23.9143 
ms per step:106.06  epoch:21/100  step:2/25  Dloss:0.2235  Gloss:23.5166 
ms per step:106.08  epoch:21/100  step:4/25  Dloss:0.3671  Gloss:23.3758 
ms per step:107.00  epoch:21/100  step:6/25  Dloss:0.2277  Gloss:23.7860 
ms per step:111.38  epoch:21/100  step:8/25  Dloss:0.2986  Gloss:24.1286 
ms per step:145.41  epoch:21/100  step:10/25  Dloss:0.2126  Gloss:24.9463 
ms per step:155.08  epoch:21/100  step:12/25  Dloss:0.4170  Gloss:20.5937 
ms per step:145.67  epoch:21/100  step:14/25  Dloss:0.1705  Gloss:22.0651 
ms per step:110.08  epoch:21/100  step:16/25  Dloss:0.1604  Gloss:23.1511 
ms per step:105.61  epoch:21/100  step:18/25  Dloss:0.1992  Gloss:22.6401 
ms per step:107.22  epoch:21/100  step:20/25  Dloss:0.2130  Gloss:24.3708 
ms per step:106.88  epoch:21/100  step:22/25  Dloss:0.1574  Gloss:26.2599 
ms per step:107.16  epoch:21/100  step:24/25  Dloss:0.2495  Gloss:23.6248 
ms per step:106.06  epoch:22/100  step:0/25  Dloss:0.2879  Gloss:23.0740 
ms per step:107.99  epoch:22/100  step:2/25  Dloss:0.2362  Gloss:25.2399 
ms per step:109.37  epoch:22/100  step:4/25  Dloss:0.3611  Gloss:22.4944 
ms per step:109.48  epoch:22/100  step:6/25  Dloss:0.1763  Gloss:24.2273 
ms per step:107.45  epoch:22/100  step:8/25  Dloss:0.1354  Gloss:25.5558 
ms per step:150.88  epoch:22/100  step:10/25  Dloss:0.4557  Gloss:25.0474 
ms per step:124.16  epoch:22/100  step:12/25  Dloss:0.1416  Gloss:24.7038 
ms per step:141.37  epoch:22/100  step:14/25  Dloss:0.2144  Gloss:24.5439 
ms per step:112.18  epoch:22/100  step:16/25  Dloss:0.2571  Gloss:21.0980 
ms per step:107.39  epoch:22/100  step:18/25  Dloss:0.3910  Gloss:22.7223 
ms per step:109.88  epoch:22/100  step:20/25  Dloss:0.4821  Gloss:23.1425 
ms per step:107.32  epoch:22/100  step:22/25  Dloss:0.3530  Gloss:23.3428 
ms per step:131.98  epoch:22/100  step:24/25  Dloss:0.1457  Gloss:24.1709 
ms per step:107.23  epoch:23/100  step:0/25  Dloss:0.2751  Gloss:21.7048 
ms per step:105.82  epoch:23/100  step:2/25  Dloss:0.2921  Gloss:20.3530 
ms per step:135.46  epoch:23/100  step:4/25  Dloss:0.2665  Gloss:20.7926 
ms per step:117.90  epoch:23/100  step:6/25  Dloss:0.2724  Gloss:21.5450 
ms per step:129.55  epoch:23/100  step:8/25  Dloss:0.1904  Gloss:21.7360 
ms per step:148.38  epoch:23/100  step:10/25  Dloss:0.3040  Gloss:25.2340 
ms per step:157.44  epoch:23/100  step:12/25  Dloss:0.3371  Gloss:21.3238 
ms per step:148.18  epoch:23/100  step:14/25  Dloss:0.3259  Gloss:23.6548 
ms per step:112.67  epoch:23/100  step:16/25  Dloss:0.4190  Gloss:22.5656 
ms per step:109.69  epoch:23/100  step:18/25  Dloss:0.3652  Gloss:30.2355 
ms per step:119.71  epoch:23/100  step:20/25  Dloss:0.2824  Gloss:24.4302 
ms per step:106.29  epoch:23/100  step:22/25  Dloss:0.3743  Gloss:20.1809 
ms per step:107.00  epoch:23/100  step:24/25  Dloss:0.2729  Gloss:21.7424 
ms per step:105.27  epoch:24/100  step:0/25  Dloss:0.2781  Gloss:20.8999 
ms per step:106.17  epoch:24/100  step:2/25  Dloss:0.3746  Gloss:20.4720 
ms per step:113.64  epoch:24/100  step:4/25  Dloss:0.3610  Gloss:20.9705 
ms per step:106.86  epoch:24/100  step:6/25  Dloss:0.1568  Gloss:23.6787 
ms per step:106.80  epoch:24/100  step:8/25  Dloss:0.1742  Gloss:20.8628 
ms per step:121.54  epoch:24/100  step:10/25  Dloss:0.2986  Gloss:20.7585 
ms per step:124.39  epoch:24/100  step:12/25  Dloss:0.3298  Gloss:19.5449 
ms per step:144.50  epoch:24/100  step:14/25  Dloss:0.7190  Gloss:20.5618 
ms per step:112.32  epoch:24/100  step:16/25  Dloss:0.3819  Gloss:20.5090 
ms per step:109.80  epoch:24/100  step:18/25  Dloss:0.2862  Gloss:25.5357 
ms per step:108.76  epoch:24/100  step:20/25  Dloss:0.1987  Gloss:20.4038 
ms per step:116.22  epoch:24/100  step:22/25  Dloss:0.1325  Gloss:22.4773 
ms per step:125.85  epoch:24/100  step:24/25  Dloss:0.2102  Gloss:21.4302 
ms per step:135.29  epoch:25/100  step:0/25  Dloss:0.2338  Gloss:20.3356 
ms per step:108.27  epoch:25/100  step:2/25  Dloss:0.2542  Gloss:18.7456 
ms per step:109.94  epoch:25/100  step:4/25  Dloss:0.2538  Gloss:23.3988 
ms per step:105.46  epoch:25/100  step:6/25  Dloss:0.2688  Gloss:20.0536 
ms per step:107.33  epoch:25/100  step:8/25  Dloss:0.4805  Gloss:23.2786 
ms per step:139.00  epoch:25/100  step:10/25  Dloss:0.4271  Gloss:19.5007 
ms per step:146.20  epoch:25/100  step:12/25  Dloss:0.4810  Gloss:21.0232 
ms per step:158.40  epoch:25/100  step:14/25  Dloss:0.2606  Gloss:21.1494 
ms per step:113.16  epoch:25/100  step:16/25  Dloss:0.1751  Gloss:21.9246 
ms per step:106.23  epoch:25/100  step:18/25  Dloss:0.1847  Gloss:21.1607 
ms per step:107.99  epoch:25/100  step:20/25  Dloss:0.1819  Gloss:20.2439 
ms per step:109.67  epoch:25/100  step:22/25  Dloss:0.2381  Gloss:22.7654 
ms per step:109.93  epoch:25/100  step:24/25  Dloss:0.4142  Gloss:18.6555 
ms per step:109.81  epoch:26/100  step:0/25  Dloss:0.4442  Gloss:18.2933 
ms per step:115.37  epoch:26/100  step:2/25  Dloss:0.3686  Gloss:18.8838 
ms per step:112.83  epoch:26/100  step:4/25  Dloss:0.1920  Gloss:21.8669 
ms per step:115.69  epoch:26/100  step:6/25  Dloss:0.2021  Gloss:20.8979 
ms per step:109.81  epoch:26/100  step:8/25  Dloss:0.1659  Gloss:22.1169 
ms per step:156.42  epoch:26/100  step:10/25  Dloss:0.4690  Gloss:19.3535 
ms per step:144.08  epoch:26/100  step:12/25  Dloss:0.3853  Gloss:21.8180 
ms per step:146.60  epoch:26/100  step:14/25  Dloss:0.8582  Gloss:23.1792 
ms per step:145.29  epoch:26/100  step:16/25  Dloss:0.3972  Gloss:18.5702 
ms per step:105.71  epoch:26/100  step:18/25  Dloss:0.3009  Gloss:18.0110 
ms per step:113.97  epoch:26/100  step:20/25  Dloss:0.1982  Gloss:22.1415 
ms per step:104.94  epoch:26/100  step:22/25  Dloss:0.2368  Gloss:19.1651 
ms per step:103.54  epoch:26/100  step:24/25  Dloss:0.1345  Gloss:21.4350 
ms per step:104.73  epoch:27/100  step:0/25  Dloss:0.1930  Gloss:18.7527 
ms per step:115.34  epoch:27/100  step:2/25  Dloss:0.3141  Gloss:19.4527 
ms per step:103.81  epoch:27/100  step:4/25  Dloss:0.2596  Gloss:18.8677 
ms per step:137.30  epoch:27/100  step:6/25  Dloss:0.4001  Gloss:20.6983 
ms per step:106.33  epoch:27/100  step:8/25  Dloss:0.2148  Gloss:20.9220 
ms per step:144.01  epoch:27/100  step:10/25  Dloss:0.2687  Gloss:17.7448 
ms per step:160.82  epoch:27/100  step:12/25  Dloss:0.3151  Gloss:17.5041 
ms per step:115.42  epoch:27/100  step:14/25  Dloss:0.1501  Gloss:20.9493 
ms per step:132.04  epoch:27/100  step:16/25  Dloss:0.1722  Gloss:18.4998 
ms per step:105.80  epoch:27/100  step:18/25  Dloss:0.2183  Gloss:19.3421 
ms per step:105.88  epoch:27/100  step:20/25  Dloss:0.2117  Gloss:18.6341 
ms per step:110.59  epoch:27/100  step:22/25  Dloss:0.3359  Gloss:18.6644 
ms per step:104.71  epoch:27/100  step:24/25  Dloss:0.7957  Gloss:21.9831 
ms per step:104.67  epoch:28/100  step:0/25  Dloss:1.0866  Gloss:20.1007 
ms per step:104.95  epoch:28/100  step:2/25  Dloss:0.2559  Gloss:20.0169 
ms per step:108.22  epoch:28/100  step:4/25  Dloss:0.3900  Gloss:18.2745 
ms per step:105.57  epoch:28/100  step:6/25  Dloss:0.4240  Gloss:20.6456 
ms per step:126.93  epoch:28/100  step:8/25  Dloss:0.3940  Gloss:19.8655 
ms per step:151.86  epoch:28/100  step:10/25  Dloss:0.3842  Gloss:17.5028 
ms per step:120.17  epoch:28/100  step:12/25  Dloss:0.2199  Gloss:19.2721 
ms per step:130.35  epoch:28/100  step:14/25  Dloss:0.2378  Gloss:19.7780 
ms per step:100.73  epoch:28/100  step:16/25  Dloss:0.2630  Gloss:18.0365 
ms per step:97.35  epoch:28/100  step:18/25  Dloss:0.4417  Gloss:17.6489 
ms per step:97.63  epoch:28/100  step:20/25  Dloss:0.3447  Gloss:17.7328 
ms per step:97.47  epoch:28/100  step:22/25  Dloss:0.2281  Gloss:19.0124 
ms per step:96.63  epoch:28/100  step:24/25  Dloss:0.3365  Gloss:21.2767 
ms per step:96.69  epoch:29/100  step:0/25  Dloss:0.3501  Gloss:18.8439 
ms per step:97.70  epoch:29/100  step:2/25  Dloss:0.1753  Gloss:17.5204 
ms per step:96.50  epoch:29/100  step:4/25  Dloss:0.1180  Gloss:21.5239 
ms per step:101.27  epoch:29/100  step:6/25  Dloss:0.1844  Gloss:17.3276 
ms per step:100.82  epoch:29/100  step:8/25  Dloss:0.2104  Gloss:19.1877 
ms per step:109.41  epoch:29/100  step:10/25  Dloss:0.3102  Gloss:19.8071 
ms per step:111.29  epoch:29/100  step:12/25  Dloss:0.3283  Gloss:18.7604 
ms per step:105.91  epoch:29/100  step:14/25  Dloss:0.3761  Gloss:19.6189 
ms per step:128.20  epoch:29/100  step:16/25  Dloss:0.4078  Gloss:17.3511 
ms per step:99.12  epoch:29/100  step:18/25  Dloss:0.2064  Gloss:17.7340 
ms per step:97.59  epoch:29/100  step:20/25  Dloss:0.1378  Gloss:17.9424 
ms per step:96.99  epoch:29/100  step:22/25  Dloss:0.1534  Gloss:17.1386 
ms per step:100.01  epoch:29/100  step:24/25  Dloss:0.1198  Gloss:18.1769 
ms per step:96.71  epoch:30/100  step:0/25  Dloss:0.1499  Gloss:17.5272 
ms per step:96.42  epoch:30/100  step:2/25  Dloss:0.2267  Gloss:18.4305 
ms per step:97.52  epoch:30/100  step:4/25  Dloss:0.7221  Gloss:17.5034 
ms per step:97.38  epoch:30/100  step:6/25  Dloss:0.3298  Gloss:16.9349 
ms per step:96.48  epoch:30/100  step:8/25  Dloss:0.3057  Gloss:15.6239 
ms per step:105.22  epoch:30/100  step:10/25  Dloss:0.2766  Gloss:16.5417 
ms per step:109.19  epoch:30/100  step:12/25  Dloss:0.2776  Gloss:16.9973 
ms per step:106.38  epoch:30/100  step:14/25  Dloss:0.3216  Gloss:18.8120 
ms per step:118.66  epoch:30/100  step:16/25  Dloss:0.2904  Gloss:17.1248 
ms per step:96.77  epoch:30/100  step:18/25  Dloss:0.3097  Gloss:18.0417 
ms per step:97.33  epoch:30/100  step:20/25  Dloss:0.1950  Gloss:17.6192 
ms per step:96.68  epoch:30/100  step:22/25  Dloss:0.4548  Gloss:16.7320 
ms per step:96.68  epoch:30/100  step:24/25  Dloss:0.2114  Gloss:19.2640 
ms per step:97.28  epoch:31/100  step:0/25  Dloss:0.3883  Gloss:16.7202 
ms per step:96.58  epoch:31/100  step:2/25  Dloss:0.3661  Gloss:20.4554 
ms per step:96.39  epoch:31/100  step:4/25  Dloss:0.1620  Gloss:19.9417 
ms per step:96.81  epoch:31/100  step:6/25  Dloss:0.3559  Gloss:18.4848 
ms per step:98.02  epoch:31/100  step:8/25  Dloss:0.2191  Gloss:17.3360 
ms per step:130.18  epoch:31/100  step:10/25  Dloss:0.2818  Gloss:16.3180 
ms per step:114.04  epoch:31/100  step:12/25  Dloss:0.3184  Gloss:16.2521 
ms per step:104.49  epoch:31/100  step:14/25  Dloss:0.4552  Gloss:18.1078 
ms per step:106.10  epoch:31/100  step:16/25  Dloss:0.4582  Gloss:15.3059 
ms per step:98.56  epoch:31/100  step:18/25  Dloss:0.4000  Gloss:16.1711 
ms per step:97.65  epoch:31/100  step:20/25  Dloss:0.2066  Gloss:16.5775 
ms per step:102.48  epoch:31/100  step:22/25  Dloss:0.2979  Gloss:17.3286 
ms per step:97.75  epoch:31/100  step:24/25  Dloss:0.2257  Gloss:16.9297 
ms per step:97.03  epoch:32/100  step:0/25  Dloss:0.2910  Gloss:16.3272 
ms per step:96.90  epoch:32/100  step:2/25  Dloss:0.4548  Gloss:21.5088 
ms per step:95.97  epoch:32/100  step:4/25  Dloss:0.2637  Gloss:17.1083 
ms per step:95.88  epoch:32/100  step:6/25  Dloss:0.3128  Gloss:16.7091 
ms per step:96.58  epoch:32/100  step:8/25  Dloss:0.2581  Gloss:16.3464 
ms per step:103.91  epoch:32/100  step:10/25  Dloss:0.5005  Gloss:15.9861 
ms per step:106.04  epoch:32/100  step:12/25  Dloss:0.4440  Gloss:17.2590 
ms per step:106.62  epoch:32/100  step:14/25  Dloss:0.2370  Gloss:16.6737 
ms per step:111.68  epoch:32/100  step:16/25  Dloss:0.1941  Gloss:15.1668 
ms per step:98.03  epoch:32/100  step:18/25  Dloss:0.1818  Gloss:17.0351 
ms per step:96.82  epoch:32/100  step:20/25  Dloss:0.2495  Gloss:15.4847 
ms per step:96.42  epoch:32/100  step:22/25  Dloss:0.2207  Gloss:16.4981 
ms per step:97.41  epoch:32/100  step:24/25  Dloss:0.2997  Gloss:15.8290 
ms per step:95.90  epoch:33/100  step:0/25  Dloss:0.4188  Gloss:14.7267 
ms per step:97.80  epoch:33/100  step:2/25  Dloss:0.3564  Gloss:16.7468 
ms per step:97.43  epoch:33/100  step:4/25  Dloss:0.2642  Gloss:17.0630 
ms per step:99.74  epoch:33/100  step:6/25  Dloss:0.3512  Gloss:17.0884 
ms per step:96.72  epoch:33/100  step:8/25  Dloss:0.2911  Gloss:16.4367 
ms per step:111.66  epoch:33/100  step:10/25  Dloss:0.3139  Gloss:16.3974 
ms per step:108.81  epoch:33/100  step:12/25  Dloss:0.1941  Gloss:16.2110 
ms per step:105.40  epoch:33/100  step:14/25  Dloss:0.2710  Gloss:18.0199 
ms per step:104.96  epoch:33/100  step:16/25  Dloss:0.5882  Gloss:16.4506 
ms per step:99.89  epoch:33/100  step:18/25  Dloss:0.5652  Gloss:14.9657 
ms per step:97.34  epoch:33/100  step:20/25  Dloss:0.3347  Gloss:16.5341 
ms per step:98.17  epoch:33/100  step:22/25  Dloss:0.2739  Gloss:16.6867 
ms per step:96.66  epoch:33/100  step:24/25  Dloss:0.2046  Gloss:17.0413 
ms per step:96.54  epoch:34/100  step:0/25  Dloss:0.1508  Gloss:16.1241 
ms per step:96.50  epoch:34/100  step:2/25  Dloss:0.2339  Gloss:13.5640 
ms per step:96.04  epoch:34/100  step:4/25  Dloss:0.2181  Gloss:17.3504 
ms per step:97.40  epoch:34/100  step:6/25  Dloss:0.5518  Gloss:15.0297 
ms per step:96.96  epoch:34/100  step:8/25  Dloss:0.4233  Gloss:16.3343 
ms per step:103.60  epoch:34/100  step:10/25  Dloss:0.3041  Gloss:15.7255 
ms per step:128.23  epoch:34/100  step:12/25  Dloss:0.2994  Gloss:15.9517 
ms per step:103.36  epoch:34/100  step:14/25  Dloss:0.3098  Gloss:14.9478 
ms per step:126.48  epoch:34/100  step:16/25  Dloss:0.2625  Gloss:16.0261 
ms per step:98.90  epoch:34/100  step:18/25  Dloss:0.5825  Gloss:16.1667 
ms per step:97.76  epoch:34/100  step:20/25  Dloss:0.5257  Gloss:15.5318 
ms per step:97.73  epoch:34/100  step:22/25  Dloss:0.7086  Gloss:17.2141 
ms per step:101.68  epoch:34/100  step:24/25  Dloss:0.3330  Gloss:16.7725 
ms per step:96.30  epoch:35/100  step:0/25  Dloss:0.3603  Gloss:15.8449 
ms per step:101.27  epoch:35/100  step:2/25  Dloss:0.2219  Gloss:16.0918 
ms per step:95.76  epoch:35/100  step:4/25  Dloss:0.4089  Gloss:16.0361 
ms per step:96.60  epoch:35/100  step:6/25  Dloss:0.3502  Gloss:16.9639 
ms per step:98.02  epoch:35/100  step:8/25  Dloss:0.3991  Gloss:15.7897 
ms per step:104.94  epoch:35/100  step:10/25  Dloss:0.2450  Gloss:15.1704 
ms per step:109.84  epoch:35/100  step:12/25  Dloss:0.3833  Gloss:14.8413 
ms per step:106.89  epoch:35/100  step:14/25  Dloss:0.1805  Gloss:14.5697 
ms per step:109.18  epoch:35/100  step:16/25  Dloss:0.2484  Gloss:14.7032 
ms per step:96.79  epoch:35/100  step:18/25  Dloss:0.2053  Gloss:16.2399 
ms per step:97.28  epoch:35/100  step:20/25  Dloss:0.3633  Gloss:14.9062 
ms per step:97.22  epoch:35/100  step:22/25  Dloss:0.4150  Gloss:15.4078 
ms per step:99.72  epoch:35/100  step:24/25  Dloss:0.2037  Gloss:16.7491 
ms per step:96.56  epoch:36/100  step:0/25  Dloss:0.3851  Gloss:14.2723 
ms per step:97.28  epoch:36/100  step:2/25  Dloss:0.2257  Gloss:16.8603 
ms per step:96.26  epoch:36/100  step:4/25  Dloss:0.3138  Gloss:15.3638 
ms per step:96.58  epoch:36/100  step:6/25  Dloss:0.2694  Gloss:15.1657 
ms per step:100.73  epoch:36/100  step:8/25  Dloss:0.5680  Gloss:15.9754 
ms per step:143.54  epoch:36/100  step:10/25  Dloss:0.3387  Gloss:17.9910 
ms per step:103.72  epoch:36/100  step:12/25  Dloss:0.7280  Gloss:14.3611 
ms per step:106.62  epoch:36/100  step:14/25  Dloss:0.3644  Gloss:15.6079 
ms per step:106.44  epoch:36/100  step:16/25  Dloss:0.4030  Gloss:13.8805 
ms per step:98.01  epoch:36/100  step:18/25  Dloss:0.4007  Gloss:14.3715 
ms per step:98.03  epoch:36/100  step:20/25  Dloss:0.3061  Gloss:17.1476 
ms per step:95.91  epoch:36/100  step:22/25  Dloss:0.3463  Gloss:17.9034 
ms per step:118.11  epoch:36/100  step:24/25  Dloss:0.2396  Gloss:15.2886 
ms per step:98.30  epoch:37/100  step:0/25  Dloss:0.5822  Gloss:15.3606 
ms per step:97.77  epoch:37/100  step:2/25  Dloss:0.2864  Gloss:14.9943 
ms per step:101.28  epoch:37/100  step:4/25  Dloss:0.4345  Gloss:15.4389 
ms per step:95.34  epoch:37/100  step:6/25  Dloss:0.4290  Gloss:14.6737 
ms per step:99.81  epoch:37/100  step:8/25  Dloss:0.4440  Gloss:13.8994 
ms per step:107.71  epoch:37/100  step:10/25  Dloss:0.3487  Gloss:13.1993 
ms per step:110.99  epoch:37/100  step:12/25  Dloss:0.3297  Gloss:15.1093 
ms per step:130.89  epoch:37/100  step:14/25  Dloss:0.1877  Gloss:15.6246 
ms per step:107.19  epoch:37/100  step:16/25  Dloss:0.1978  Gloss:14.8823 
ms per step:100.42  epoch:37/100  step:18/25  Dloss:0.3243  Gloss:14.6991 
ms per step:96.94  epoch:37/100  step:20/25  Dloss:0.3623  Gloss:13.9333 
ms per step:97.56  epoch:37/100  step:22/25  Dloss:0.8251  Gloss:15.5981 
ms per step:98.40  epoch:37/100  step:24/25  Dloss:0.3141  Gloss:14.6158 
ms per step:112.60  epoch:38/100  step:0/25  Dloss:0.2940  Gloss:14.3617 
ms per step:107.98  epoch:38/100  step:2/25  Dloss:0.2311  Gloss:15.4742 
ms per step:101.60  epoch:38/100  step:4/25  Dloss:0.2054  Gloss:14.2449 
ms per step:97.68  epoch:38/100  step:6/25  Dloss:0.1813  Gloss:13.1655 
ms per step:99.64  epoch:38/100  step:8/25  Dloss:0.2529  Gloss:13.8493 
ms per step:139.75  epoch:38/100  step:10/25  Dloss:0.5853  Gloss:15.7463 
ms per step:140.51  epoch:38/100  step:12/25  Dloss:0.6124  Gloss:15.2568 
ms per step:129.07  epoch:38/100  step:14/25  Dloss:0.4089  Gloss:14.1301 
ms per step:102.14  epoch:38/100  step:16/25  Dloss:0.2702  Gloss:13.6872 
ms per step:101.78  epoch:38/100  step:18/25  Dloss:0.2379  Gloss:14.7659 
ms per step:102.41  epoch:38/100  step:20/25  Dloss:0.2659  Gloss:15.7267 
ms per step:101.50  epoch:38/100  step:22/25  Dloss:0.1935  Gloss:16.8655 
ms per step:100.05  epoch:38/100  step:24/25  Dloss:0.2827  Gloss:15.6609 
ms per step:99.65  epoch:39/100  step:0/25  Dloss:0.4028  Gloss:12.8613 
ms per step:98.20  epoch:39/100  step:2/25  Dloss:0.3947  Gloss:15.7700 
ms per step:101.94  epoch:39/100  step:4/25  Dloss:0.3869  Gloss:14.1484 
ms per step:98.92  epoch:39/100  step:6/25  Dloss:0.2303  Gloss:14.7239 
ms per step:106.24  epoch:39/100  step:8/25  Dloss:0.2705  Gloss:14.9515 
ms per step:106.33  epoch:39/100  step:10/25  Dloss:0.2508  Gloss:14.1135 
ms per step:106.70  epoch:39/100  step:12/25  Dloss:0.4355  Gloss:14.8306 
ms per step:110.41  epoch:39/100  step:14/25  Dloss:0.4513  Gloss:14.5726 
ms per step:135.72  epoch:39/100  step:16/25  Dloss:0.6247  Gloss:15.3168 
ms per step:98.45  epoch:39/100  step:18/25  Dloss:0.3027  Gloss:14.8332 
ms per step:99.15  epoch:39/100  step:20/25  Dloss:0.2331  Gloss:15.1761 
ms per step:98.64  epoch:39/100  step:22/25  Dloss:0.2657  Gloss:14.6750 
ms per step:98.76  epoch:39/100  step:24/25  Dloss:0.4145  Gloss:14.2217 
ms per step:98.15  epoch:40/100  step:0/25  Dloss:0.7223  Gloss:13.7114 
ms per step:97.60  epoch:40/100  step:2/25  Dloss:0.8602  Gloss:13.8505 
ms per step:98.88  epoch:40/100  step:4/25  Dloss:0.4073  Gloss:15.1188 
ms per step:97.34  epoch:40/100  step:6/25  Dloss:0.3739  Gloss:15.1431 
ms per step:98.03  epoch:40/100  step:8/25  Dloss:0.2511  Gloss:14.2960 
ms per step:127.32  epoch:40/100  step:10/25  Dloss:0.2506  Gloss:14.3230 
ms per step:106.12  epoch:40/100  step:12/25  Dloss:0.3550  Gloss:17.5432 
ms per step:105.25  epoch:40/100  step:14/25  Dloss:0.1936  Gloss:16.7018 
ms per step:105.45  epoch:40/100  step:16/25  Dloss:0.2809  Gloss:13.4992 
ms per step:98.29  epoch:40/100  step:18/25  Dloss:0.3368  Gloss:12.2888 
ms per step:99.62  epoch:40/100  step:20/25  Dloss:0.3178  Gloss:14.7571 
ms per step:97.46  epoch:40/100  step:22/25  Dloss:0.2947  Gloss:13.9504 
ms per step:97.73  epoch:40/100  step:24/25  Dloss:0.3001  Gloss:13.5222 
ms per step:98.75  epoch:41/100  step:0/25  Dloss:0.2473  Gloss:14.4535 
ms per step:97.16  epoch:41/100  step:2/25  Dloss:0.2101  Gloss:15.0425 
ms per step:98.30  epoch:41/100  step:4/25  Dloss:0.2087  Gloss:15.7633 
ms per step:97.57  epoch:41/100  step:6/25  Dloss:0.4985  Gloss:13.6644 
ms per step:98.06  epoch:41/100  step:8/25  Dloss:0.5168  Gloss:14.9434 
ms per step:105.78  epoch:41/100  step:10/25  Dloss:0.5632  Gloss:14.8322 
ms per step:105.74  epoch:41/100  step:12/25  Dloss:0.3882  Gloss:14.5167 
ms per step:125.86  epoch:41/100  step:14/25  Dloss:0.3156  Gloss:13.4330 
ms per step:108.53  epoch:41/100  step:16/25  Dloss:0.2077  Gloss:15.2035 
ms per step:98.00  epoch:41/100  step:18/25  Dloss:0.2644  Gloss:13.5223 
ms per step:98.52  epoch:41/100  step:20/25  Dloss:0.2473  Gloss:13.8799 
ms per step:99.64  epoch:41/100  step:22/25  Dloss:0.3658  Gloss:11.9632 
ms per step:97.83  epoch:41/100  step:24/25  Dloss:0.5015  Gloss:13.5292 
ms per step:99.21  epoch:42/100  step:0/25  Dloss:0.5054  Gloss:13.9852 
ms per step:97.32  epoch:42/100  step:2/25  Dloss:0.3696  Gloss:13.3376 
ms per step:100.23  epoch:42/100  step:4/25  Dloss:0.2582  Gloss:13.5459 
ms per step:97.76  epoch:42/100  step:6/25  Dloss:0.2201  Gloss:12.5249 
ms per step:98.61  epoch:42/100  step:8/25  Dloss:0.2892  Gloss:12.2936 
ms per step:105.34  epoch:42/100  step:10/25  Dloss:0.2926  Gloss:13.2115 
ms per step:106.55  epoch:42/100  step:12/25  Dloss:0.8180  Gloss:15.4382 
ms per step:108.12  epoch:42/100  step:14/25  Dloss:0.5467  Gloss:14.1481 
ms per step:105.28  epoch:42/100  step:16/25  Dloss:0.3230  Gloss:13.8746 
ms per step:99.18  epoch:42/100  step:18/25  Dloss:0.3606  Gloss:14.1825 
ms per step:97.81  epoch:42/100  step:20/25  Dloss:0.2653  Gloss:13.1178 
ms per step:99.04  epoch:42/100  step:22/25  Dloss:0.3527  Gloss:16.1646 
ms per step:98.61  epoch:42/100  step:24/25  Dloss:0.3772  Gloss:15.8270 
ms per step:99.15  epoch:43/100  step:0/25  Dloss:0.3760  Gloss:13.8284 
ms per step:100.23  epoch:43/100  step:2/25  Dloss:0.5371  Gloss:12.5458 
ms per step:105.79  epoch:43/100  step:4/25  Dloss:0.3896  Gloss:14.1019 
ms per step:100.85  epoch:43/100  step:6/25  Dloss:0.3092  Gloss:14.4113 
ms per step:104.89  epoch:43/100  step:8/25  Dloss:0.3022  Gloss:12.4935 
ms per step:135.75  epoch:43/100  step:10/25  Dloss:0.2667  Gloss:13.8905 
ms per step:106.41  epoch:43/100  step:12/25  Dloss:0.6177  Gloss:12.8958 
ms per step:103.78  epoch:43/100  step:14/25  Dloss:0.5342  Gloss:13.3103 
ms per step:105.20  epoch:43/100  step:16/25  Dloss:0.4589  Gloss:13.0687 
ms per step:101.37  epoch:43/100  step:18/25  Dloss:0.3227  Gloss:16.5795 
ms per step:96.69  epoch:43/100  step:20/25  Dloss:0.4688  Gloss:13.0339 
ms per step:99.33  epoch:43/100  step:22/25  Dloss:0.4984  Gloss:12.2296 
ms per step:98.08  epoch:43/100  step:24/25  Dloss:0.4598  Gloss:14.5670 
ms per step:96.34  epoch:44/100  step:0/25  Dloss:0.8077  Gloss:14.2064 
ms per step:97.27  epoch:44/100  step:2/25  Dloss:0.5531  Gloss:12.7416 
ms per step:96.36  epoch:44/100  step:4/25  Dloss:0.4489  Gloss:13.7550 
ms per step:98.28  epoch:44/100  step:6/25  Dloss:0.5393  Gloss:12.6501 
ms per step:98.07  epoch:44/100  step:8/25  Dloss:0.3647  Gloss:13.5786 
ms per step:107.02  epoch:44/100  step:10/25  Dloss:0.3748  Gloss:13.9286 
ms per step:104.73  epoch:44/100  step:12/25  Dloss:0.3769  Gloss:14.9645 
ms per step:104.83  epoch:44/100  step:14/25  Dloss:0.3898  Gloss:13.8285 
ms per step:119.18  epoch:44/100  step:16/25  Dloss:0.4265  Gloss:13.5696 
ms per step:97.16  epoch:44/100  step:18/25  Dloss:0.5924  Gloss:12.6545 
ms per step:97.23  epoch:44/100  step:20/25  Dloss:0.4313  Gloss:12.8075 
ms per step:96.39  epoch:44/100  step:22/25  Dloss:0.3763  Gloss:12.5626 
ms per step:96.83  epoch:44/100  step:24/25  Dloss:0.4616  Gloss:11.5262 
ms per step:97.44  epoch:45/100  step:0/25  Dloss:0.4459  Gloss:14.1252 
ms per step:96.61  epoch:45/100  step:2/25  Dloss:0.2761  Gloss:13.4958 
ms per step:96.92  epoch:45/100  step:4/25  Dloss:0.5254  Gloss:12.6856 
ms per step:96.94  epoch:45/100  step:6/25  Dloss:0.4519  Gloss:12.6776 
ms per step:97.35  epoch:45/100  step:8/25  Dloss:0.7917  Gloss:13.1139 
ms per step:126.68  epoch:45/100  step:10/25  Dloss:0.5488  Gloss:12.7250 
ms per step:105.11  epoch:45/100  step:12/25  Dloss:0.4481  Gloss:12.8375 
ms per step:106.17  epoch:45/100  step:14/25  Dloss:0.4875  Gloss:12.3398 
ms per step:132.50  epoch:45/100  step:16/25  Dloss:0.3656  Gloss:12.2982 
ms per step:97.33  epoch:45/100  step:18/25  Dloss:0.3069  Gloss:12.6054 
ms per step:95.55  epoch:45/100  step:20/25  Dloss:0.3213  Gloss:12.9775 
ms per step:97.07  epoch:45/100  step:22/25  Dloss:0.4016  Gloss:11.3416 
ms per step:96.62  epoch:45/100  step:24/25  Dloss:0.3987  Gloss:12.4437 
ms per step:108.90  epoch:46/100  step:0/25  Dloss:0.3723  Gloss:12.7536 
ms per step:97.98  epoch:46/100  step:2/25  Dloss:0.5461  Gloss:14.9380 
ms per step:95.82  epoch:46/100  step:4/25  Dloss:0.2856  Gloss:13.2420 
ms per step:96.09  epoch:46/100  step:6/25  Dloss:0.2898  Gloss:12.3241 
ms per step:96.97  epoch:46/100  step:8/25  Dloss:0.5355  Gloss:12.8218 
ms per step:105.72  epoch:46/100  step:10/25  Dloss:0.7188  Gloss:13.1113 
ms per step:106.38  epoch:46/100  step:12/25  Dloss:0.5568  Gloss:13.1963 
ms per step:107.95  epoch:46/100  step:14/25  Dloss:0.4144  Gloss:12.8916 
ms per step:104.67  epoch:46/100  step:16/25  Dloss:0.3715  Gloss:11.9469 
ms per step:96.00  epoch:46/100  step:18/25  Dloss:0.3513  Gloss:12.4033 
ms per step:96.96  epoch:46/100  step:20/25  Dloss:0.3673  Gloss:13.0759 
ms per step:106.88  epoch:46/100  step:22/25  Dloss:0.5264  Gloss:12.4882 
ms per step:101.91  epoch:46/100  step:24/25  Dloss:0.6276  Gloss:11.7067 
ms per step:97.86  epoch:47/100  step:0/25  Dloss:0.8157  Gloss:14.1273 
ms per step:97.61  epoch:47/100  step:2/25  Dloss:0.3069  Gloss:13.6492 
ms per step:98.55  epoch:47/100  step:4/25  Dloss:0.3247  Gloss:12.2684 
ms per step:96.82  epoch:47/100  step:6/25  Dloss:0.2620  Gloss:13.1911 
ms per step:97.44  epoch:47/100  step:8/25  Dloss:0.3609  Gloss:13.2669 
ms per step:103.90  epoch:47/100  step:10/25  Dloss:0.7484  Gloss:13.2896 
ms per step:105.86  epoch:47/100  step:12/25  Dloss:0.5265  Gloss:13.4602 
ms per step:104.97  epoch:47/100  step:14/25  Dloss:0.5921  Gloss:12.4433 
ms per step:121.51  epoch:47/100  step:16/25  Dloss:0.4571  Gloss:12.3026 
ms per step:96.06  epoch:47/100  step:18/25  Dloss:0.3974  Gloss:11.7008 
ms per step:97.47  epoch:47/100  step:20/25  Dloss:0.3674  Gloss:12.4367 
ms per step:102.13  epoch:47/100  step:22/25  Dloss:0.2900  Gloss:12.4297 
ms per step:101.64  epoch:47/100  step:24/25  Dloss:0.4196  Gloss:10.8767 
ms per step:97.30  epoch:48/100  step:0/25  Dloss:0.4383  Gloss:10.7088 
ms per step:100.98  epoch:48/100  step:2/25  Dloss:0.4166  Gloss:13.5078 
ms per step:100.49  epoch:48/100  step:4/25  Dloss:0.5590  Gloss:11.7251 
ms per step:98.61  epoch:48/100  step:6/25  Dloss:0.6213  Gloss:11.8545 
ms per step:98.95  epoch:48/100  step:8/25  Dloss:0.3619  Gloss:11.7317 
ms per step:105.81  epoch:48/100  step:10/25  Dloss:0.4362  Gloss:12.9007 
ms per step:107.01  epoch:48/100  step:12/25  Dloss:0.2797  Gloss:12.7246 
ms per step:104.52  epoch:48/100  step:14/25  Dloss:0.4781  Gloss:11.5047 
ms per step:115.96  epoch:48/100  step:16/25  Dloss:0.6003  Gloss:11.5007 
ms per step:100.01  epoch:48/100  step:18/25  Dloss:0.4442  Gloss:12.2953 
ms per step:99.22  epoch:48/100  step:20/25  Dloss:0.5811  Gloss:12.2399 
ms per step:97.80  epoch:48/100  step:22/25  Dloss:0.4685  Gloss:12.3062 
ms per step:101.98  epoch:48/100  step:24/25  Dloss:0.5672  Gloss:12.7262 
ms per step:97.31  epoch:49/100  step:0/25  Dloss:0.5458  Gloss:14.0043 
ms per step:97.26  epoch:49/100  step:2/25  Dloss:0.2912  Gloss:12.3867 
ms per step:96.60  epoch:49/100  step:4/25  Dloss:0.3920  Gloss:13.0228 
ms per step:101.64  epoch:49/100  step:6/25  Dloss:0.6817  Gloss:13.0263 
ms per step:96.77  epoch:49/100  step:8/25  Dloss:0.6422  Gloss:11.8432 
ms per step:104.06  epoch:49/100  step:10/25  Dloss:0.5683  Gloss:12.0917 
ms per step:117.51  epoch:49/100  step:12/25  Dloss:0.3371  Gloss:12.7969 
ms per step:110.67  epoch:49/100  step:14/25  Dloss:0.4230  Gloss:11.2478 
ms per step:111.12  epoch:49/100  step:16/25  Dloss:0.3103  Gloss:11.3678 
ms per step:101.08  epoch:49/100  step:18/25  Dloss:0.4252  Gloss:12.0863 
ms per step:96.92  epoch:49/100  step:20/25  Dloss:0.4168  Gloss:12.8535 
ms per step:98.02  epoch:49/100  step:22/25  Dloss:0.7757  Gloss:11.7664 
ms per step:97.42  epoch:49/100  step:24/25  Dloss:0.5229  Gloss:13.2126 
ms per step:97.67  epoch:50/100  step:0/25  Dloss:0.5795  Gloss:12.2427 
ms per step:96.35  epoch:50/100  step:2/25  Dloss:0.3731  Gloss:12.0884 
ms per step:99.31  epoch:50/100  step:4/25  Dloss:0.3703  Gloss:11.1971 
ms per step:103.30  epoch:50/100  step:6/25  Dloss:0.3482  Gloss:12.9322 
ms per step:100.52  epoch:50/100  step:8/25  Dloss:0.4353  Gloss:12.5127 
ms per step:106.95  epoch:50/100  step:10/25  Dloss:0.5801  Gloss:11.3761 
ms per step:119.65  epoch:50/100  step:12/25  Dloss:0.3625  Gloss:11.7094 
ms per step:134.82  epoch:50/100  step:14/25  Dloss:0.5007  Gloss:12.0538 
ms per step:108.04  epoch:50/100  step:16/25  Dloss:0.5092  Gloss:12.1419 
ms per step:99.83  epoch:50/100  step:18/25  Dloss:0.6654  Gloss:11.6835 
ms per step:100.93  epoch:50/100  step:20/25  Dloss:0.3819  Gloss:12.3631 
ms per step:104.23  epoch:50/100  step:22/25  Dloss:0.3556  Gloss:11.9356 
ms per step:99.91  epoch:50/100  step:24/25  Dloss:0.3005  Gloss:11.9859 
ms per step:100.95  epoch:51/100  step:0/25  Dloss:0.2768  Gloss:11.7612 
ms per step:99.77  epoch:51/100  step:2/25  Dloss:0.5767  Gloss:11.7958 
ms per step:99.35  epoch:51/100  step:4/25  Dloss:0.6627  Gloss:11.8194 
ms per step:101.07  epoch:51/100  step:6/25  Dloss:0.7372  Gloss:11.3262 
ms per step:99.29  epoch:51/100  step:8/25  Dloss:0.4244  Gloss:12.5333 
ms per step:108.02  epoch:51/100  step:10/25  Dloss:0.3954  Gloss:11.1332 
ms per step:106.32  epoch:51/100  step:12/25  Dloss:0.3181  Gloss:11.9812 
ms per step:107.33  epoch:51/100  step:14/25  Dloss:0.2828  Gloss:11.4513 
ms per step:104.24  epoch:51/100  step:16/25  Dloss:0.4032  Gloss:12.0914 
ms per step:97.96  epoch:51/100  step:18/25  Dloss:0.4253  Gloss:12.0994 
ms per step:98.07  epoch:51/100  step:20/25  Dloss:0.4757  Gloss:13.8628 
ms per step:96.75  epoch:51/100  step:22/25  Dloss:0.4933  Gloss:11.4924 
ms per step:96.71  epoch:51/100  step:24/25  Dloss:0.3689  Gloss:12.6341 
ms per step:97.12  epoch:52/100  step:0/25  Dloss:0.3674  Gloss:12.0886 
ms per step:100.29  epoch:52/100  step:2/25  Dloss:0.3505  Gloss:10.7783 
ms per step:96.45  epoch:52/100  step:4/25  Dloss:0.4006  Gloss:12.5076 
ms per step:97.44  epoch:52/100  step:6/25  Dloss:0.3684  Gloss:12.9985 
ms per step:100.46  epoch:52/100  step:8/25  Dloss:0.4971  Gloss:11.6595 
ms per step:106.68  epoch:52/100  step:10/25  Dloss:0.8337  Gloss:13.3431 
ms per step:105.35  epoch:52/100  step:12/25  Dloss:0.4074  Gloss:11.3735 
ms per step:104.86  epoch:52/100  step:14/25  Dloss:0.4305  Gloss:11.6835 
ms per step:105.58  epoch:52/100  step:16/25  Dloss:0.5156  Gloss:11.4562 
ms per step:98.04  epoch:52/100  step:18/25  Dloss:0.6212  Gloss:11.4215 
ms per step:100.43  epoch:52/100  step:20/25  Dloss:0.3508  Gloss:13.1385 
ms per step:97.51  epoch:52/100  step:22/25  Dloss:0.3570  Gloss:11.8557 
ms per step:97.45  epoch:52/100  step:24/25  Dloss:0.4029  Gloss:11.9170 
ms per step:97.04  epoch:53/100  step:0/25  Dloss:0.4193  Gloss:10.9841 
ms per step:98.54  epoch:53/100  step:2/25  Dloss:0.3667  Gloss:12.0504 
ms per step:97.63  epoch:53/100  step:4/25  Dloss:0.5321  Gloss:13.2151 
ms per step:98.98  epoch:53/100  step:6/25  Dloss:0.5907  Gloss:11.9479 
ms per step:97.90  epoch:53/100  step:8/25  Dloss:0.4285  Gloss:11.9451 
ms per step:105.47  epoch:53/100  step:10/25  Dloss:0.2677  Gloss:12.7194 
ms per step:105.37  epoch:53/100  step:12/25  Dloss:0.2697  Gloss:11.3483 
ms per step:105.60  epoch:53/100  step:14/25  Dloss:0.3215  Gloss:11.4588 
ms per step:107.64  epoch:53/100  step:16/25  Dloss:0.4777  Gloss:12.1538 
ms per step:97.42  epoch:53/100  step:18/25  Dloss:0.8412  Gloss:13.3639 
ms per step:98.07  epoch:53/100  step:20/25  Dloss:0.7679  Gloss:11.9476 
ms per step:95.86  epoch:53/100  step:22/25  Dloss:0.4972  Gloss:12.5294 
ms per step:96.95  epoch:53/100  step:24/25  Dloss:0.3818  Gloss:11.2317 
ms per step:96.02  epoch:54/100  step:0/25  Dloss:0.3488  Gloss:12.7709 
ms per step:97.36  epoch:54/100  step:2/25  Dloss:0.2984  Gloss:12.1337 
ms per step:97.03  epoch:54/100  step:4/25  Dloss:0.5031  Gloss:10.2937 
ms per step:96.95  epoch:54/100  step:6/25  Dloss:0.5138  Gloss:10.4693 
ms per step:96.68  epoch:54/100  step:8/25  Dloss:0.4432  Gloss:11.4550 
ms per step:105.67  epoch:54/100  step:10/25  Dloss:0.4123  Gloss:12.3023 
ms per step:105.79  epoch:54/100  step:12/25  Dloss:0.4407  Gloss:11.1429 
ms per step:105.08  epoch:54/100  step:14/25  Dloss:0.3392  Gloss:11.7460 
ms per step:107.09  epoch:54/100  step:16/25  Dloss:0.3291  Gloss:11.0476 
ms per step:100.47  epoch:54/100  step:18/25  Dloss:0.4755  Gloss:11.1438 
ms per step:98.54  epoch:54/100  step:20/25  Dloss:0.3497  Gloss:11.1534 
ms per step:100.46  epoch:54/100  step:22/25  Dloss:0.3545  Gloss:11.6293 
ms per step:98.02  epoch:54/100  step:24/25  Dloss:0.4226  Gloss:13.0790 
ms per step:98.10  epoch:55/100  step:0/25  Dloss:0.4310  Gloss:11.3384 
ms per step:97.43  epoch:55/100  step:2/25  Dloss:0.4633  Gloss:10.7768 
ms per step:97.95  epoch:55/100  step:4/25  Dloss:0.3709  Gloss:11.9606 
ms per step:99.28  epoch:55/100  step:6/25  Dloss:0.4031  Gloss:11.2025 
ms per step:97.27  epoch:55/100  step:8/25  Dloss:0.5278  Gloss:11.8508 
ms per step:104.06  epoch:55/100  step:10/25  Dloss:0.5248  Gloss:11.8630 
ms per step:104.53  epoch:55/100  step:12/25  Dloss:0.6093  Gloss:12.1857 
ms per step:103.06  epoch:55/100  step:14/25  Dloss:0.3471  Gloss:12.1212 
ms per step:103.18  epoch:55/100  step:16/25  Dloss:0.4096  Gloss:12.1174 
ms per step:97.54  epoch:55/100  step:18/25  Dloss:0.5687  Gloss:11.3148 
ms per step:97.44  epoch:55/100  step:20/25  Dloss:0.5130  Gloss:11.4408 
ms per step:105.66  epoch:55/100  step:22/25  Dloss:0.5215  Gloss:11.8262 
ms per step:100.71  epoch:55/100  step:24/25  Dloss:0.5147  Gloss:11.5154 
ms per step:97.40  epoch:56/100  step:0/25  Dloss:0.4542  Gloss:10.5607 
ms per step:98.20  epoch:56/100  step:2/25  Dloss:0.3198  Gloss:12.7340 
ms per step:97.62  epoch:56/100  step:4/25  Dloss:0.4362  Gloss:12.6521 
ms per step:97.11  epoch:56/100  step:6/25  Dloss:0.2941  Gloss:11.1877 
ms per step:97.08  epoch:56/100  step:8/25  Dloss:0.3534  Gloss:10.9625 
ms per step:103.18  epoch:56/100  step:10/25  Dloss:0.3315  Gloss:11.5538 
ms per step:102.90  epoch:56/100  step:12/25  Dloss:0.4579  Gloss:12.8862 
ms per step:103.25  epoch:56/100  step:14/25  Dloss:0.2711  Gloss:12.2871 
ms per step:105.80  epoch:56/100  step:16/25  Dloss:0.4124  Gloss:11.0175 
ms per step:100.82  epoch:56/100  step:18/25  Dloss:0.3928  Gloss:11.3780 
ms per step:97.52  epoch:56/100  step:20/25  Dloss:0.3673  Gloss:11.2428 
ms per step:100.53  epoch:56/100  step:22/25  Dloss:0.4246  Gloss:11.2912 
ms per step:100.54  epoch:56/100  step:24/25  Dloss:0.3636  Gloss:11.5546 
ms per step:97.86  epoch:57/100  step:0/25  Dloss:0.3135  Gloss:12.6024 
ms per step:97.02  epoch:57/100  step:2/25  Dloss:0.2846  Gloss:12.7947 
ms per step:98.53  epoch:57/100  step:4/25  Dloss:0.1950  Gloss:13.1210 
ms per step:97.19  epoch:57/100  step:6/25  Dloss:0.4789  Gloss:12.1808 
ms per step:98.77  epoch:57/100  step:8/25  Dloss:1.0249  Gloss:11.2005 
ms per step:105.06  epoch:57/100  step:10/25  Dloss:0.5618  Gloss:12.2865 
ms per step:105.16  epoch:57/100  step:12/25  Dloss:0.4195  Gloss:11.5979 
ms per step:107.21  epoch:57/100  step:14/25  Dloss:0.2279  Gloss:11.9391 
ms per step:105.54  epoch:57/100  step:16/25  Dloss:0.2636  Gloss:12.2098 
ms per step:95.31  epoch:57/100  step:18/25  Dloss:0.4779  Gloss:12.2958 
ms per step:95.60  epoch:57/100  step:20/25  Dloss:0.6658  Gloss:12.0602 
ms per step:97.24  epoch:57/100  step:22/25  Dloss:0.7232  Gloss:12.4160 
ms per step:94.89  epoch:57/100  step:24/25  Dloss:0.3936  Gloss:11.1692 
ms per step:96.56  epoch:58/100  step:0/25  Dloss:0.4153  Gloss:10.0797 
ms per step:96.52  epoch:58/100  step:2/25  Dloss:0.4698  Gloss:10.8337 
ms per step:95.51  epoch:58/100  step:4/25  Dloss:0.7222  Gloss:11.4187 
ms per step:95.96  epoch:58/100  step:6/25  Dloss:0.5471  Gloss:12.5028 
ms per step:98.09  epoch:58/100  step:8/25  Dloss:0.5436  Gloss:10.6078 
ms per step:123.89  epoch:58/100  step:10/25  Dloss:0.3751  Gloss:10.7369 
ms per step:104.70  epoch:58/100  step:12/25  Dloss:0.3950  Gloss:10.9655 
ms per step:104.99  epoch:58/100  step:14/25  Dloss:0.3142  Gloss:11.3345 
ms per step:105.56  epoch:58/100  step:16/25  Dloss:0.2803  Gloss:11.4462 
ms per step:97.42  epoch:58/100  step:18/25  Dloss:0.3864  Gloss:10.9695 
ms per step:97.40  epoch:58/100  step:20/25  Dloss:0.5120  Gloss:11.7873 
ms per step:97.50  epoch:58/100  step:22/25  Dloss:0.4303  Gloss:10.8452 
ms per step:96.77  epoch:58/100  step:24/25  Dloss:0.4576  Gloss:10.7594 
ms per step:96.20  epoch:59/100  step:0/25  Dloss:0.2809  Gloss:12.4908 
ms per step:97.94  epoch:59/100  step:2/25  Dloss:0.4187  Gloss:11.1622 
ms per step:96.26  epoch:59/100  step:4/25  Dloss:0.3375  Gloss:11.2609 
ms per step:96.20  epoch:59/100  step:6/25  Dloss:0.3192  Gloss:10.9699 
ms per step:97.05  epoch:59/100  step:8/25  Dloss:0.2934  Gloss:11.1831 
ms per step:105.21  epoch:59/100  step:10/25  Dloss:0.3769  Gloss:13.4488 
ms per step:103.70  epoch:59/100  step:12/25  Dloss:0.3773  Gloss:11.7507 
ms per step:139.62  epoch:59/100  step:14/25  Dloss:0.2886  Gloss:11.8666 
ms per step:111.05  epoch:59/100  step:16/25  Dloss:0.2437  Gloss:12.5939 
ms per step:96.58  epoch:59/100  step:18/25  Dloss:0.4620  Gloss:11.3511 
ms per step:98.33  epoch:59/100  step:20/25  Dloss:0.7487  Gloss:10.5775 
ms per step:101.42  epoch:59/100  step:22/25  Dloss:0.4970  Gloss:14.6681 
ms per step:97.38  epoch:59/100  step:24/25  Dloss:0.3267  Gloss:11.2459 
ms per step:104.91  epoch:60/100  step:0/25  Dloss:0.3448  Gloss:11.3164 
ms per step:96.43  epoch:60/100  step:2/25  Dloss:0.3652  Gloss:10.4771 
ms per step:96.51  epoch:60/100  step:4/25  Dloss:0.5229  Gloss:10.8432 
ms per step:97.19  epoch:60/100  step:6/25  Dloss:0.4796  Gloss:13.1254 
ms per step:95.40  epoch:60/100  step:8/25  Dloss:0.6358  Gloss:12.9104 
ms per step:117.81  epoch:60/100  step:10/25  Dloss:0.4124  Gloss:11.6393 
ms per step:112.62  epoch:60/100  step:12/25  Dloss:0.4005  Gloss:10.8148 
ms per step:106.44  epoch:60/100  step:14/25  Dloss:0.4766  Gloss:10.9539 
ms per step:116.96  epoch:60/100  step:16/25  Dloss:0.4589  Gloss:9.6456 
ms per step:98.36  epoch:60/100  step:18/25  Dloss:0.5077  Gloss:10.9015 
ms per step:99.44  epoch:60/100  step:20/25  Dloss:0.3087  Gloss:11.1825 
ms per step:96.82  epoch:60/100  step:22/25  Dloss:0.3830  Gloss:11.9578 
ms per step:97.30  epoch:60/100  step:24/25  Dloss:0.2887  Gloss:12.6586 
ms per step:100.26  epoch:61/100  step:0/25  Dloss:0.3832  Gloss:11.3697 
ms per step:100.00  epoch:61/100  step:2/25  Dloss:0.3762  Gloss:11.2534 
ms per step:102.41  epoch:61/100  step:4/25  Dloss:0.3569  Gloss:11.0815 
ms per step:100.03  epoch:61/100  step:6/25  Dloss:0.3109  Gloss:10.9940 
ms per step:97.14  epoch:61/100  step:8/25  Dloss:0.2491  Gloss:11.1991 
ms per step:135.14  epoch:61/100  step:10/25  Dloss:0.2806  Gloss:10.0623 
ms per step:106.83  epoch:61/100  step:12/25  Dloss:0.2456  Gloss:11.3681 
ms per step:134.62  epoch:61/100  step:14/25  Dloss:0.2156  Gloss:10.7684 
ms per step:103.08  epoch:61/100  step:16/25  Dloss:0.3906  Gloss:11.2433 
ms per step:100.48  epoch:61/100  step:18/25  Dloss:0.9109  Gloss:10.9749 
ms per step:95.47  epoch:61/100  step:20/25  Dloss:0.5805  Gloss:12.0455 
ms per step:96.95  epoch:61/100  step:22/25  Dloss:0.3869  Gloss:9.9103 
ms per step:95.59  epoch:61/100  step:24/25  Dloss:0.3435  Gloss:10.6366 
ms per step:97.84  epoch:62/100  step:0/25  Dloss:0.3136  Gloss:10.3077 
ms per step:95.54  epoch:62/100  step:2/25  Dloss:0.3806  Gloss:9.5601 
ms per step:98.72  epoch:62/100  step:4/25  Dloss:0.4390  Gloss:9.9382 
ms per step:96.04  epoch:62/100  step:6/25  Dloss:0.4904  Gloss:10.8363 
ms per step:98.72  epoch:62/100  step:8/25  Dloss:0.5896  Gloss:10.2382 
ms per step:103.72  epoch:62/100  step:10/25  Dloss:0.5475  Gloss:10.2236 
ms per step:104.88  epoch:62/100  step:12/25  Dloss:0.4197  Gloss:9.6912 
ms per step:126.90  epoch:62/100  step:14/25  Dloss:0.2477  Gloss:10.6298 
ms per step:104.92  epoch:62/100  step:16/25  Dloss:0.2212  Gloss:10.2353 
ms per step:96.66  epoch:62/100  step:18/25  Dloss:0.2580  Gloss:10.5120 
ms per step:96.77  epoch:62/100  step:20/25  Dloss:0.2491  Gloss:10.9265 
ms per step:97.32  epoch:62/100  step:22/25  Dloss:0.2607  Gloss:10.7973 
ms per step:95.70  epoch:62/100  step:24/25  Dloss:0.3067  Gloss:12.3965 
ms per step:112.41  epoch:63/100  step:0/25  Dloss:0.2871  Gloss:11.0337 
ms per step:95.48  epoch:63/100  step:2/25  Dloss:0.5242  Gloss:10.8298 
ms per step:96.20  epoch:63/100  step:4/25  Dloss:0.6841  Gloss:10.2130 
ms per step:96.77  epoch:63/100  step:6/25  Dloss:0.3946  Gloss:11.1338 
ms per step:98.42  epoch:63/100  step:8/25  Dloss:0.2881  Gloss:11.2007 
ms per step:130.04  epoch:63/100  step:10/25  Dloss:0.2628  Gloss:10.8245 
ms per step:131.72  epoch:63/100  step:12/25  Dloss:0.2894  Gloss:10.2425 
ms per step:106.34  epoch:63/100  step:14/25  Dloss:0.4948  Gloss:11.4727 
ms per step:126.67  epoch:63/100  step:16/25  Dloss:0.5281  Gloss:10.6207 
ms per step:96.15  epoch:63/100  step:18/25  Dloss:0.3243  Gloss:10.6962 
ms per step:95.76  epoch:63/100  step:20/25  Dloss:0.3467  Gloss:9.6237 
ms per step:101.15  epoch:63/100  step:22/25  Dloss:0.3269  Gloss:10.4317 
ms per step:97.84  epoch:63/100  step:24/25  Dloss:0.5747  Gloss:10.8460 
ms per step:96.83  epoch:64/100  step:0/25  Dloss:0.5359  Gloss:9.5159 
ms per step:96.60  epoch:64/100  step:2/25  Dloss:0.3563  Gloss:10.8629 
ms per step:95.59  epoch:64/100  step:4/25  Dloss:0.2906  Gloss:10.3898 
ms per step:97.83  epoch:64/100  step:6/25  Dloss:0.2906  Gloss:10.1712 
ms per step:95.21  epoch:64/100  step:8/25  Dloss:0.3272  Gloss:10.4990 
ms per step:120.87  epoch:64/100  step:10/25  Dloss:0.4307  Gloss:12.1158 
ms per step:102.03  epoch:64/100  step:12/25  Dloss:0.8277  Gloss:10.1580 
ms per step:98.41  epoch:64/100  step:14/25  Dloss:0.4002  Gloss:12.2934 
ms per step:99.78  epoch:64/100  step:16/25  Dloss:0.5348  Gloss:9.7776 
ms per step:93.82  epoch:64/100  step:18/25  Dloss:0.4108  Gloss:10.9043 
ms per step:94.38  epoch:64/100  step:20/25  Dloss:0.4693  Gloss:9.6580 
ms per step:93.85  epoch:64/100  step:22/25  Dloss:0.4357  Gloss:10.8924 
ms per step:94.60  epoch:64/100  step:24/25  Dloss:0.4319  Gloss:10.8629 
ms per step:93.27  epoch:65/100  step:0/25  Dloss:0.3080  Gloss:10.6347 
ms per step:94.37  epoch:65/100  step:2/25  Dloss:0.3609  Gloss:10.1289 
ms per step:94.04  epoch:65/100  step:4/25  Dloss:0.4060  Gloss:9.7318 
ms per step:94.33  epoch:65/100  step:6/25  Dloss:0.4503  Gloss:10.0633 
ms per step:94.54  epoch:65/100  step:8/25  Dloss:0.4129  Gloss:10.1854 
ms per step:103.14  epoch:65/100  step:10/25  Dloss:0.4164  Gloss:9.8382 
ms per step:103.32  epoch:65/100  step:12/25  Dloss:0.2852  Gloss:11.2488 
ms per step:126.55  epoch:65/100  step:14/25  Dloss:0.2692  Gloss:11.3541 
ms per step:103.07  epoch:65/100  step:16/25  Dloss:0.2951  Gloss:11.4776 
ms per step:98.30  epoch:65/100  step:18/25  Dloss:0.3812  Gloss:10.0599 
ms per step:98.18  epoch:65/100  step:20/25  Dloss:0.3825  Gloss:10.1102 
ms per step:98.26  epoch:65/100  step:22/25  Dloss:0.2820  Gloss:12.0682 
ms per step:98.59  epoch:65/100  step:24/25  Dloss:0.3236  Gloss:10.8278 
ms per step:98.15  epoch:66/100  step:0/25  Dloss:0.4436  Gloss:9.5112 
ms per step:97.53  epoch:66/100  step:2/25  Dloss:0.6358  Gloss:10.2870 
ms per step:98.02  epoch:66/100  step:4/25  Dloss:0.5921  Gloss:10.0417 
ms per step:98.59  epoch:66/100  step:6/25  Dloss:0.3965  Gloss:10.0804 
ms per step:97.91  epoch:66/100  step:8/25  Dloss:0.3912  Gloss:9.9749 
ms per step:105.38  epoch:66/100  step:10/25  Dloss:0.3262  Gloss:10.9261 
ms per step:104.97  epoch:66/100  step:12/25  Dloss:0.3048  Gloss:11.8741 
ms per step:105.59  epoch:66/100  step:14/25  Dloss:0.3400  Gloss:10.1248 
ms per step:105.43  epoch:66/100  step:16/25  Dloss:0.4324  Gloss:10.6249 
ms per step:99.85  epoch:66/100  step:18/25  Dloss:0.5531  Gloss:10.2458 
ms per step:99.83  epoch:66/100  step:20/25  Dloss:0.5239  Gloss:11.1489 
ms per step:99.70  epoch:66/100  step:22/25  Dloss:0.4444  Gloss:10.8603 
ms per step:104.29  epoch:66/100  step:24/25  Dloss:0.3388  Gloss:10.2709 
ms per step:98.11  epoch:67/100  step:0/25  Dloss:0.5258  Gloss:10.1396 
ms per step:97.73  epoch:67/100  step:2/25  Dloss:0.4375  Gloss:11.0381 
ms per step:99.01  epoch:67/100  step:4/25  Dloss:0.4873  Gloss:10.1493 
ms per step:104.88  epoch:67/100  step:6/25  Dloss:0.5326  Gloss:9.9929 
ms per step:103.57  epoch:67/100  step:8/25  Dloss:0.5997  Gloss:9.1982 
ms per step:121.43  epoch:67/100  step:10/25  Dloss:0.3866  Gloss:12.0144 
ms per step:107.92  epoch:67/100  step:12/25  Dloss:0.3841  Gloss:9.5720 
ms per step:104.81  epoch:67/100  step:14/25  Dloss:0.4080  Gloss:10.1461 
ms per step:103.20  epoch:67/100  step:16/25  Dloss:0.2889  Gloss:9.8612 
ms per step:96.74  epoch:67/100  step:18/25  Dloss:0.2261  Gloss:12.4217 
ms per step:97.94  epoch:67/100  step:20/25  Dloss:0.3116  Gloss:11.0826 
ms per step:97.79  epoch:67/100  step:22/25  Dloss:0.3531  Gloss:11.3827 
ms per step:102.96  epoch:67/100  step:24/25  Dloss:0.3600  Gloss:9.9141 
ms per step:99.96  epoch:68/100  step:0/25  Dloss:0.3486  Gloss:10.8720 
ms per step:97.52  epoch:68/100  step:2/25  Dloss:0.4211  Gloss:10.1502 
ms per step:96.71  epoch:68/100  step:4/25  Dloss:0.4289  Gloss:9.7176 
ms per step:96.65  epoch:68/100  step:6/25  Dloss:0.2981  Gloss:10.2061 
ms per step:97.47  epoch:68/100  step:8/25  Dloss:0.2923  Gloss:10.2384 
ms per step:121.56  epoch:68/100  step:10/25  Dloss:0.4898  Gloss:10.8271 
ms per step:103.25  epoch:68/100  step:12/25  Dloss:0.4413  Gloss:11.7350 
ms per step:112.28  epoch:68/100  step:14/25  Dloss:0.3606  Gloss:10.3653 
ms per step:112.31  epoch:68/100  step:16/25  Dloss:0.3274  Gloss:12.0149 
ms per step:111.80  epoch:68/100  step:18/25  Dloss:0.2556  Gloss:10.6520 
ms per step:111.65  epoch:68/100  step:20/25  Dloss:0.3072  Gloss:10.7027 
ms per step:112.14  epoch:68/100  step:22/25  Dloss:0.3712  Gloss:9.9511 
ms per step:111.61  epoch:68/100  step:24/25  Dloss:0.3914  Gloss:10.1926 
ms per step:110.93  epoch:69/100  step:0/25  Dloss:0.3120  Gloss:10.7071 
ms per step:111.46  epoch:69/100  step:2/25  Dloss:0.2765  Gloss:11.9192 
ms per step:111.76  epoch:69/100  step:4/25  Dloss:0.6389  Gloss:10.4836 
ms per step:113.68  epoch:69/100  step:6/25  Dloss:0.6278  Gloss:9.6649 
ms per step:113.53  epoch:69/100  step:8/25  Dloss:0.5052  Gloss:10.2801 
ms per step:113.64  epoch:69/100  step:10/25  Dloss:0.3726  Gloss:10.4907 
ms per step:114.46  epoch:69/100  step:12/25  Dloss:0.3663  Gloss:11.2954 
ms per step:111.95  epoch:69/100  step:14/25  Dloss:0.4155  Gloss:10.5536 
ms per step:112.16  epoch:69/100  step:16/25  Dloss:0.4200  Gloss:10.1128 
ms per step:112.88  epoch:69/100  step:18/25  Dloss:0.3785  Gloss:10.2815 
ms per step:111.22  epoch:69/100  step:20/25  Dloss:0.3550  Gloss:10.1851 
ms per step:112.35  epoch:69/100  step:22/25  Dloss:0.2829  Gloss:10.3639 
ms per step:111.65  epoch:69/100  step:24/25  Dloss:0.2831  Gloss:10.9097 
ms per step:111.14  epoch:70/100  step:0/25  Dloss:0.3712  Gloss:9.6637 
ms per step:111.43  epoch:70/100  step:2/25  Dloss:0.3807  Gloss:11.0406 
ms per step:111.42  epoch:70/100  step:4/25  Dloss:0.2961  Gloss:9.7218 
ms per step:112.72  epoch:70/100  step:6/25  Dloss:0.3864  Gloss:10.8362 
ms per step:110.67  epoch:70/100  step:8/25  Dloss:0.6103  Gloss:10.2481 
ms per step:112.17  epoch:70/100  step:10/25  Dloss:0.4058  Gloss:9.8044 
ms per step:111.24  epoch:70/100  step:12/25  Dloss:0.3210  Gloss:11.3604 
ms per step:112.28  epoch:70/100  step:14/25  Dloss:0.3232  Gloss:9.9605 
ms per step:111.45  epoch:70/100  step:16/25  Dloss:0.4388  Gloss:9.5607 
ms per step:111.71  epoch:70/100  step:18/25  Dloss:0.3801  Gloss:10.5030 
ms per step:111.56  epoch:70/100  step:20/25  Dloss:0.2966  Gloss:10.2318 
ms per step:112.76  epoch:70/100  step:22/25  Dloss:0.3002  Gloss:11.1848 
ms per step:112.84  epoch:70/100  step:24/25  Dloss:0.3088  Gloss:9.8187 
ms per step:112.43  epoch:71/100  step:0/25  Dloss:0.3538  Gloss:11.0118 
ms per step:112.82  epoch:71/100  step:2/25  Dloss:0.3106  Gloss:10.8807 
ms per step:114.76  epoch:71/100  step:4/25  Dloss:0.8549  Gloss:10.5364 
ms per step:115.19  epoch:71/100  step:6/25  Dloss:0.3765  Gloss:10.2776 
ms per step:114.84  epoch:71/100  step:8/25  Dloss:0.3905  Gloss:9.7115 
ms per step:115.85  epoch:71/100  step:10/25  Dloss:0.2988  Gloss:11.0878 
ms per step:111.42  epoch:71/100  step:12/25  Dloss:0.4946  Gloss:10.1506 
ms per step:112.42  epoch:71/100  step:14/25  Dloss:0.4960  Gloss:9.9633 
ms per step:111.59  epoch:71/100  step:16/25  Dloss:0.5449  Gloss:10.2751 
ms per step:112.21  epoch:71/100  step:18/25  Dloss:0.4223  Gloss:9.7398 
ms per step:112.36  epoch:71/100  step:20/25  Dloss:0.3043  Gloss:10.6444 
ms per step:111.82  epoch:71/100  step:22/25  Dloss:0.3263  Gloss:10.0748 
ms per step:111.41  epoch:71/100  step:24/25  Dloss:0.3390  Gloss:9.7063 
ms per step:111.92  epoch:72/100  step:0/25  Dloss:0.4619  Gloss:9.6069 
ms per step:112.23  epoch:72/100  step:2/25  Dloss:0.4932  Gloss:10.3540 
ms per step:112.62  epoch:72/100  step:4/25  Dloss:0.3847  Gloss:10.5311 
ms per step:113.04  epoch:72/100  step:6/25  Dloss:0.3851  Gloss:10.2513 
ms per step:111.44  epoch:72/100  step:8/25  Dloss:0.5441  Gloss:11.1488 
ms per step:112.60  epoch:72/100  step:10/25  Dloss:0.5738  Gloss:9.7600 
ms per step:111.59  epoch:72/100  step:12/25  Dloss:0.4670  Gloss:10.6421 
ms per step:112.97  epoch:72/100  step:14/25  Dloss:0.5086  Gloss:9.4348 
ms per step:112.40  epoch:72/100  step:16/25  Dloss:0.3026  Gloss:10.3118 
ms per step:112.79  epoch:72/100  step:18/25  Dloss:0.5338  Gloss:9.6670 
ms per step:111.19  epoch:72/100  step:20/25  Dloss:0.7645  Gloss:9.2889 
ms per step:111.90  epoch:72/100  step:22/25  Dloss:0.6808  Gloss:9.9566 
ms per step:111.58  epoch:72/100  step:24/25  Dloss:0.5532  Gloss:10.4015 
ms per step:112.16  epoch:73/100  step:0/25  Dloss:0.4996  Gloss:9.9770 
ms per step:111.42  epoch:73/100  step:2/25  Dloss:0.4413  Gloss:9.9476 
ms per step:112.00  epoch:73/100  step:4/25  Dloss:0.5027  Gloss:9.6874 
ms per step:112.40  epoch:73/100  step:6/25  Dloss:0.5505  Gloss:10.0737 
ms per step:111.56  epoch:73/100  step:8/25  Dloss:0.5534  Gloss:9.5679 
ms per step:113.33  epoch:73/100  step:10/25  Dloss:0.3968  Gloss:10.0447 
ms per step:111.56  epoch:73/100  step:12/25  Dloss:0.2536  Gloss:9.9949 
ms per step:110.88  epoch:73/100  step:14/25  Dloss:0.2582  Gloss:10.2137 
ms per step:111.64  epoch:73/100  step:16/25  Dloss:0.2401  Gloss:10.7389 
ms per step:111.17  epoch:73/100  step:18/25  Dloss:0.3058  Gloss:10.3998 
ms per step:124.51  epoch:73/100  step:20/25  Dloss:0.3300  Gloss:9.2162 
ms per step:110.21  epoch:73/100  step:22/25  Dloss:0.3060  Gloss:10.1422 
ms per step:111.54  epoch:73/100  step:24/25  Dloss:0.4356  Gloss:9.8226 
ms per step:111.31  epoch:74/100  step:0/25  Dloss:0.4134  Gloss:11.4354 
ms per step:111.58  epoch:74/100  step:2/25  Dloss:0.5282  Gloss:11.6185 
ms per step:111.64  epoch:74/100  step:4/25  Dloss:0.5384  Gloss:10.6329 
ms per step:111.57  epoch:74/100  step:6/25  Dloss:0.3445  Gloss:10.6945 
ms per step:110.93  epoch:74/100  step:8/25  Dloss:0.3406  Gloss:9.4997 
ms per step:111.73  epoch:74/100  step:10/25  Dloss:0.3698  Gloss:9.7121 
ms per step:110.27  epoch:74/100  step:12/25  Dloss:0.4388  Gloss:10.2939 
ms per step:110.73  epoch:74/100  step:14/25  Dloss:0.3199  Gloss:10.0521 
ms per step:110.61  epoch:74/100  step:16/25  Dloss:0.3250  Gloss:10.9757 
ms per step:110.03  epoch:74/100  step:18/25  Dloss:0.2886  Gloss:10.3067 
ms per step:111.62  epoch:74/100  step:20/25  Dloss:0.3874  Gloss:9.9333 
ms per step:113.19  epoch:74/100  step:22/25  Dloss:0.5159  Gloss:10.2885 
ms per step:113.12  epoch:74/100  step:24/25  Dloss:0.8759  Gloss:10.2262 
ms per step:113.26  epoch:75/100  step:0/25  Dloss:0.6922  Gloss:11.1287 
ms per step:111.76  epoch:75/100  step:2/25  Dloss:0.4278  Gloss:10.1106 
ms per step:111.76  epoch:75/100  step:4/25  Dloss:0.3800  Gloss:9.7830 
ms per step:111.43  epoch:75/100  step:6/25  Dloss:0.3365  Gloss:10.3153 
ms per step:112.49  epoch:75/100  step:8/25  Dloss:0.4983  Gloss:12.7500 
ms per step:111.86  epoch:75/100  step:10/25  Dloss:0.5590  Gloss:9.9113 
ms per step:112.02  epoch:75/100  step:12/25  Dloss:0.5224  Gloss:9.8311 
ms per step:112.36  epoch:75/100  step:14/25  Dloss:0.3127  Gloss:11.4523 
ms per step:113.19  epoch:75/100  step:16/25  Dloss:0.2899  Gloss:10.0867 
ms per step:113.29  epoch:75/100  step:18/25  Dloss:0.2013  Gloss:10.2714 
ms per step:112.32  epoch:75/100  step:20/25  Dloss:0.3988  Gloss:10.7119 
ms per step:112.18  epoch:75/100  step:22/25  Dloss:0.5196  Gloss:8.9916 
ms per step:111.33  epoch:75/100  step:24/25  Dloss:0.4304  Gloss:9.6492 
ms per step:112.17  epoch:76/100  step:0/25  Dloss:0.2842  Gloss:9.7472 
ms per step:111.65  epoch:76/100  step:2/25  Dloss:0.2529  Gloss:10.6721 
ms per step:112.44  epoch:76/100  step:4/25  Dloss:0.3277  Gloss:11.1472 
ms per step:112.20  epoch:76/100  step:6/25  Dloss:0.3904  Gloss:9.3650 
ms per step:112.96  epoch:76/100  step:8/25  Dloss:0.3469  Gloss:10.3913 
ms per step:111.85  epoch:76/100  step:10/25  Dloss:0.2060  Gloss:10.7349 
ms per step:112.29  epoch:76/100  step:12/25  Dloss:0.2695  Gloss:10.6654 
ms per step:112.07  epoch:76/100  step:14/25  Dloss:0.4983  Gloss:9.9978 
ms per step:112.23  epoch:76/100  step:16/25  Dloss:0.5894  Gloss:10.4508 
ms per step:115.19  epoch:76/100  step:18/25  Dloss:0.3223  Gloss:10.7518 
ms per step:115.01  epoch:76/100  step:20/25  Dloss:0.2871  Gloss:9.8419 
ms per step:115.21  epoch:76/100  step:22/25  Dloss:0.4450  Gloss:9.5872 
ms per step:113.28  epoch:76/100  step:24/25  Dloss:0.4459  Gloss:10.4134 
ms per step:112.89  epoch:77/100  step:0/25  Dloss:0.6214  Gloss:11.0200 
ms per step:111.46  epoch:77/100  step:2/25  Dloss:0.3538  Gloss:11.6674 
ms per step:111.79  epoch:77/100  step:4/25  Dloss:0.4930  Gloss:11.1073 
ms per step:111.67  epoch:77/100  step:6/25  Dloss:0.4974  Gloss:10.0012 
ms per step:111.43  epoch:77/100  step:8/25  Dloss:0.2700  Gloss:10.0305 
ms per step:111.14  epoch:77/100  step:10/25  Dloss:0.2416  Gloss:11.2681 
ms per step:111.37  epoch:77/100  step:12/25  Dloss:0.2444  Gloss:9.7211 
ms per step:112.92  epoch:77/100  step:14/25  Dloss:0.4125  Gloss:9.3450 
ms per step:112.30  epoch:77/100  step:16/25  Dloss:0.4492  Gloss:8.8163 
ms per step:111.68  epoch:77/100  step:18/25  Dloss:0.4431  Gloss:8.9012 
ms per step:111.80  epoch:77/100  step:20/25  Dloss:0.3819  Gloss:9.4937 
ms per step:111.55  epoch:77/100  step:22/25  Dloss:0.3244  Gloss:10.6494 
ms per step:111.16  epoch:77/100  step:24/25  Dloss:0.7647  Gloss:10.6731 
ms per step:111.70  epoch:78/100  step:0/25  Dloss:0.5483  Gloss:9.0661 
ms per step:112.23  epoch:78/100  step:2/25  Dloss:0.5922  Gloss:9.6511 
ms per step:111.20  epoch:78/100  step:4/25  Dloss:0.4281  Gloss:11.5016 
ms per step:111.92  epoch:78/100  step:6/25  Dloss:0.3651  Gloss:10.4079 
ms per step:112.22  epoch:78/100  step:8/25  Dloss:0.3027  Gloss:9.9539 
ms per step:110.91  epoch:78/100  step:10/25  Dloss:0.2464  Gloss:9.6554 
ms per step:111.01  epoch:78/100  step:12/25  Dloss:0.2238  Gloss:9.5541 
ms per step:111.18  epoch:78/100  step:14/25  Dloss:0.1853  Gloss:9.2444 
ms per step:111.54  epoch:78/100  step:16/25  Dloss:0.2081  Gloss:9.9307 
ms per step:111.87  epoch:78/100  step:18/25  Dloss:0.3129  Gloss:10.4714 
ms per step:110.80  epoch:78/100  step:20/25  Dloss:0.3006  Gloss:10.7240 
ms per step:112.41  epoch:78/100  step:22/25  Dloss:0.2183  Gloss:10.0669 
ms per step:112.15  epoch:78/100  step:24/25  Dloss:0.2740  Gloss:9.9920 
ms per step:111.84  epoch:79/100  step:0/25  Dloss:0.2426  Gloss:10.7850 
ms per step:111.26  epoch:79/100  step:2/25  Dloss:0.4743  Gloss:10.5548 
ms per step:111.05  epoch:79/100  step:4/25  Dloss:0.8711  Gloss:12.3339 
ms per step:110.82  epoch:79/100  step:6/25  Dloss:0.4723  Gloss:10.3027 
ms per step:111.03  epoch:79/100  step:8/25  Dloss:0.3755  Gloss:9.7425 
ms per step:112.22  epoch:79/100  step:10/25  Dloss:0.3536  Gloss:9.2374 
ms per step:122.59  epoch:79/100  step:12/25  Dloss:0.4414  Gloss:9.9007 
ms per step:110.90  epoch:79/100  step:14/25  Dloss:0.4690  Gloss:9.4154 
ms per step:112.08  epoch:79/100  step:16/25  Dloss:0.4539  Gloss:10.3059 
ms per step:114.24  epoch:79/100  step:18/25  Dloss:0.3346  Gloss:10.2512 
ms per step:122.56  epoch:79/100  step:20/25  Dloss:0.3908  Gloss:9.8570 
ms per step:117.47  epoch:79/100  step:22/25  Dloss:0.2460  Gloss:9.6681 
ms per step:117.53  epoch:79/100  step:24/25  Dloss:0.2672  Gloss:10.6654 
ms per step:116.22  epoch:80/100  step:0/25  Dloss:0.3916  Gloss:10.1822 
ms per step:117.14  epoch:80/100  step:2/25  Dloss:0.4663  Gloss:9.3014 
ms per step:119.41  epoch:80/100  step:4/25  Dloss:0.3364  Gloss:11.1980 
ms per step:98.69  epoch:80/100  step:6/25  Dloss:0.2556  Gloss:10.9685 
ms per step:99.76  epoch:80/100  step:8/25  Dloss:0.3316  Gloss:9.9553 
ms per step:106.58  epoch:80/100  step:10/25  Dloss:0.3088  Gloss:11.1560 
ms per step:104.32  epoch:80/100  step:12/25  Dloss:0.4410  Gloss:10.9212 
ms per step:106.49  epoch:80/100  step:14/25  Dloss:0.2633  Gloss:10.4481 
ms per step:105.40  epoch:80/100  step:16/25  Dloss:0.3358  Gloss:9.8952 
ms per step:101.08  epoch:80/100  step:18/25  Dloss:0.4061  Gloss:9.2761 
ms per step:96.76  epoch:80/100  step:20/25  Dloss:0.4931  Gloss:9.5494 
ms per step:96.94  epoch:80/100  step:22/25  Dloss:0.4216  Gloss:9.3181 
ms per step:98.28  epoch:80/100  step:24/25  Dloss:0.2979  Gloss:9.3917 
ms per step:96.64  epoch:81/100  step:0/25  Dloss:0.3254  Gloss:8.5578 
ms per step:112.01  epoch:81/100  step:2/25  Dloss:0.2435  Gloss:10.0758 
ms per step:110.70  epoch:81/100  step:4/25  Dloss:0.2373  Gloss:11.5544 
ms per step:111.66  epoch:81/100  step:6/25  Dloss:0.2365  Gloss:10.5612 
ms per step:112.11  epoch:81/100  step:8/25  Dloss:0.4334  Gloss:9.8110 
ms per step:103.20  epoch:81/100  step:10/25  Dloss:0.4046  Gloss:9.2686 
ms per step:103.37  epoch:81/100  step:12/25  Dloss:0.3842  Gloss:9.8059 
ms per step:103.44  epoch:81/100  step:14/25  Dloss:0.2537  Gloss:10.0492 
ms per step:103.40  epoch:81/100  step:16/25  Dloss:0.3426  Gloss:10.3981 
ms per step:97.66  epoch:81/100  step:18/25  Dloss:0.5250  Gloss:10.3311 
ms per step:96.33  epoch:81/100  step:20/25  Dloss:0.3806  Gloss:9.7836 
ms per step:96.84  epoch:81/100  step:22/25  Dloss:0.2444  Gloss:10.3146 
ms per step:97.20  epoch:81/100  step:24/25  Dloss:0.2971  Gloss:10.1281 
ms per step:97.98  epoch:82/100  step:0/25  Dloss:0.4627  Gloss:8.6214 
ms per step:96.33  epoch:82/100  step:2/25  Dloss:0.3178  Gloss:10.0734 
ms per step:96.86  epoch:82/100  step:4/25  Dloss:0.6950  Gloss:10.5191 
ms per step:96.69  epoch:82/100  step:6/25  Dloss:0.4393  Gloss:9.4402 
ms per step:103.70  epoch:82/100  step:8/25  Dloss:0.3814  Gloss:9.0396 
ms per step:108.10  epoch:82/100  step:10/25  Dloss:0.2999  Gloss:9.3065 
ms per step:109.13  epoch:82/100  step:12/25  Dloss:0.3494  Gloss:10.0600 
ms per step:108.16  epoch:82/100  step:14/25  Dloss:0.5465  Gloss:11.4129 
ms per step:111.67  epoch:82/100  step:16/25  Dloss:0.6140  Gloss:9.5745 
ms per step:111.10  epoch:82/100  step:18/25  Dloss:0.4674  Gloss:9.5563 
ms per step:110.31  epoch:82/100  step:20/25  Dloss:0.3460  Gloss:9.4380 
ms per step:111.88  epoch:82/100  step:22/25  Dloss:0.3767  Gloss:9.5148 
ms per step:111.83  epoch:82/100  step:24/25  Dloss:0.4124  Gloss:10.1881 
ms per step:111.58  epoch:83/100  step:0/25  Dloss:0.4318  Gloss:10.3649 
ms per step:111.19  epoch:83/100  step:2/25  Dloss:0.4005  Gloss:9.4482 
ms per step:111.20  epoch:83/100  step:4/25  Dloss:0.4203  Gloss:10.4656 
ms per step:112.31  epoch:83/100  step:6/25  Dloss:0.2092  Gloss:11.4033 
ms per step:112.11  epoch:83/100  step:8/25  Dloss:0.4368  Gloss:11.4867 
ms per step:113.54  epoch:83/100  step:10/25  Dloss:0.4125  Gloss:10.2429 
ms per step:112.94  epoch:83/100  step:12/25  Dloss:0.4689  Gloss:10.0852 
ms per step:112.96  epoch:83/100  step:14/25  Dloss:0.4591  Gloss:9.8234 
ms per step:112.47  epoch:83/100  step:16/25  Dloss:0.4539  Gloss:10.0137 
ms per step:112.92  epoch:83/100  step:18/25  Dloss:0.3868  Gloss:10.1371 
ms per step:113.74  epoch:83/100  step:20/25  Dloss:0.3291  Gloss:10.2661 
ms per step:111.80  epoch:83/100  step:22/25  Dloss:0.2847  Gloss:10.7631 
ms per step:116.49  epoch:83/100  step:24/25  Dloss:0.4388  Gloss:9.0304 
ms per step:98.09  epoch:84/100  step:0/25  Dloss:0.3811  Gloss:10.2457 
ms per step:98.95  epoch:84/100  step:2/25  Dloss:0.3117  Gloss:9.6367 
ms per step:97.07  epoch:84/100  step:4/25  Dloss:0.1935  Gloss:9.6929 
ms per step:102.96  epoch:84/100  step:6/25  Dloss:0.3382  Gloss:9.5341 
ms per step:104.37  epoch:84/100  step:8/25  Dloss:0.5542  Gloss:9.2897 
ms per step:110.86  epoch:84/100  step:10/25  Dloss:0.6423  Gloss:9.5739 
ms per step:103.04  epoch:84/100  step:12/25  Dloss:0.5439  Gloss:10.2187 
ms per step:103.47  epoch:84/100  step:14/25  Dloss:0.4691  Gloss:10.5195 
ms per step:102.74  epoch:84/100  step:16/25  Dloss:0.3856  Gloss:10.5953 
ms per step:97.92  epoch:84/100  step:18/25  Dloss:0.5081  Gloss:9.8506 
ms per step:96.77  epoch:84/100  step:20/25  Dloss:0.4532  Gloss:9.6790 
ms per step:97.48  epoch:84/100  step:22/25  Dloss:0.3681  Gloss:9.0480 
ms per step:98.72  epoch:84/100  step:24/25  Dloss:0.2787  Gloss:10.1615 
ms per step:98.09  epoch:85/100  step:0/25  Dloss:0.2226  Gloss:10.8064 
ms per step:98.46  epoch:85/100  step:2/25  Dloss:0.2190  Gloss:10.2913 
ms per step:96.77  epoch:85/100  step:4/25  Dloss:0.3360  Gloss:8.1038 
ms per step:97.44  epoch:85/100  step:6/25  Dloss:0.5133  Gloss:9.9374 
ms per step:97.29  epoch:85/100  step:8/25  Dloss:0.6348  Gloss:8.9125 
ms per step:102.85  epoch:85/100  step:10/25  Dloss:0.4057  Gloss:10.2970 
ms per step:102.54  epoch:85/100  step:12/25  Dloss:0.4176  Gloss:9.1480 
ms per step:102.58  epoch:85/100  step:14/25  Dloss:0.3760  Gloss:9.2915 
ms per step:103.10  epoch:85/100  step:16/25  Dloss:0.6374  Gloss:9.9738 
ms per step:97.23  epoch:85/100  step:18/25  Dloss:0.5142  Gloss:9.3254 
ms per step:97.31  epoch:85/100  step:20/25  Dloss:0.4496  Gloss:9.2737 
ms per step:96.60  epoch:85/100  step:22/25  Dloss:0.4027  Gloss:9.8223 
ms per step:97.64  epoch:85/100  step:24/25  Dloss:0.4317  Gloss:10.5317 
ms per step:96.93  epoch:86/100  step:0/25  Dloss:0.5546  Gloss:10.6139 
ms per step:101.15  epoch:86/100  step:2/25  Dloss:0.7265  Gloss:10.3918 
ms per step:97.68  epoch:86/100  step:4/25  Dloss:0.6294  Gloss:10.8395 
ms per step:113.81  epoch:86/100  step:6/25  Dloss:0.5798  Gloss:10.2184 
ms per step:97.24  epoch:86/100  step:8/25  Dloss:0.4173  Gloss:8.7923 
ms per step:126.92  epoch:86/100  step:10/25  Dloss:0.3636  Gloss:9.7361 
ms per step:102.60  epoch:86/100  step:12/25  Dloss:0.2965  Gloss:10.0618 
ms per step:102.48  epoch:86/100  step:14/25  Dloss:0.3613  Gloss:9.8821 
ms per step:102.11  epoch:86/100  step:16/25  Dloss:0.5689  Gloss:8.5053 
ms per step:97.25  epoch:86/100  step:18/25  Dloss:0.4054  Gloss:9.7452 
ms per step:95.95  epoch:86/100  step:20/25  Dloss:0.4335  Gloss:9.1996 
ms per step:96.91  epoch:86/100  step:22/25  Dloss:0.3853  Gloss:10.1427 
ms per step:96.66  epoch:86/100  step:24/25  Dloss:0.3381  Gloss:8.8921 
ms per step:97.49  epoch:87/100  step:0/25  Dloss:0.2935  Gloss:10.6559 
ms per step:97.17  epoch:87/100  step:2/25  Dloss:0.3589  Gloss:10.4310 
ms per step:96.56  epoch:87/100  step:4/25  Dloss:0.5330  Gloss:8.5873 
ms per step:97.11  epoch:87/100  step:6/25  Dloss:0.4251  Gloss:9.0821 
ms per step:97.10  epoch:87/100  step:8/25  Dloss:0.2924  Gloss:9.5639 
ms per step:102.99  epoch:87/100  step:10/25  Dloss:0.4142  Gloss:10.6429 
ms per step:98.28  epoch:87/100  step:12/25  Dloss:0.4027  Gloss:9.6695 
ms per step:111.54  epoch:87/100  step:14/25  Dloss:0.5504  Gloss:9.1726 
ms per step:111.38  epoch:87/100  step:16/25  Dloss:0.6566  Gloss:8.5626 
ms per step:111.60  epoch:87/100  step:18/25  Dloss:0.7008  Gloss:8.8900 
ms per step:111.31  epoch:87/100  step:20/25  Dloss:0.4265  Gloss:9.2961 
ms per step:110.92  epoch:87/100  step:22/25  Dloss:0.3628  Gloss:9.4010 
ms per step:108.55  epoch:87/100  step:24/25  Dloss:0.3666  Gloss:9.3868 
ms per step:108.61  epoch:88/100  step:0/25  Dloss:0.3866  Gloss:8.3818 
ms per step:109.78  epoch:88/100  step:2/25  Dloss:0.4337  Gloss:8.5507 
ms per step:109.50  epoch:88/100  step:4/25  Dloss:0.6346  Gloss:10.9413 
ms per step:109.84  epoch:88/100  step:6/25  Dloss:0.4474  Gloss:9.9396 
ms per step:110.41  epoch:88/100  step:8/25  Dloss:0.3321  Gloss:9.2524 
ms per step:116.23  epoch:88/100  step:10/25  Dloss:0.2697  Gloss:9.8194 
ms per step:115.25  epoch:88/100  step:12/25  Dloss:0.3174  Gloss:9.2472 
ms per step:113.67  epoch:88/100  step:14/25  Dloss:0.3723  Gloss:8.9942 
ms per step:112.36  epoch:88/100  step:16/25  Dloss:0.4940  Gloss:9.2127 
ms per step:112.88  epoch:88/100  step:18/25  Dloss:0.3799  Gloss:9.2897 
ms per step:112.39  epoch:88/100  step:20/25  Dloss:0.3532  Gloss:9.7581 
ms per step:113.19  epoch:88/100  step:22/25  Dloss:0.3137  Gloss:9.8068 
ms per step:112.45  epoch:88/100  step:24/25  Dloss:0.3797  Gloss:9.6468 
ms per step:112.32  epoch:89/100  step:0/25  Dloss:0.4515  Gloss:8.9195 
ms per step:112.55  epoch:89/100  step:2/25  Dloss:0.3397  Gloss:9.8530 
ms per step:120.04  epoch:89/100  step:4/25  Dloss:0.2947  Gloss:9.5219 
ms per step:117.49  epoch:89/100  step:6/25  Dloss:0.8607  Gloss:9.6555 
ms per step:117.80  epoch:89/100  step:8/25  Dloss:0.5896  Gloss:9.2845 
ms per step:116.03  epoch:89/100  step:10/25  Dloss:0.4360  Gloss:9.1805 
ms per step:116.45  epoch:89/100  step:12/25  Dloss:0.2512  Gloss:10.1967 
ms per step:118.49  epoch:89/100  step:14/25  Dloss:0.2471  Gloss:10.0135 
ms per step:112.87  epoch:89/100  step:16/25  Dloss:0.2895  Gloss:9.3571 
ms per step:107.90  epoch:89/100  step:18/25  Dloss:0.3593  Gloss:8.2896 
ms per step:109.69  epoch:89/100  step:20/25  Dloss:0.3980  Gloss:8.8242 
ms per step:109.80  epoch:89/100  step:22/25  Dloss:0.3117  Gloss:8.8560 
ms per step:108.83  epoch:89/100  step:24/25  Dloss:0.3242  Gloss:9.6835 
ms per step:108.99  epoch:90/100  step:0/25  Dloss:0.3349  Gloss:10.0468 
ms per step:109.92  epoch:90/100  step:2/25  Dloss:0.3614  Gloss:10.2525 
ms per step:112.68  epoch:90/100  step:4/25  Dloss:0.5712  Gloss:10.6646 
ms per step:111.89  epoch:90/100  step:6/25  Dloss:0.4600  Gloss:9.0134 
ms per step:112.62  epoch:90/100  step:8/25  Dloss:0.2493  Gloss:9.9602 
ms per step:117.01  epoch:90/100  step:10/25  Dloss:0.3217  Gloss:8.6008 
ms per step:114.07  epoch:90/100  step:12/25  Dloss:0.4104  Gloss:9.5935 
ms per step:115.49  epoch:90/100  step:14/25  Dloss:0.5306  Gloss:9.5365 
ms per step:114.34  epoch:90/100  step:16/25  Dloss:0.3813  Gloss:9.3473 
ms per step:110.82  epoch:90/100  step:18/25  Dloss:0.4325  Gloss:8.3572 
ms per step:108.99  epoch:90/100  step:20/25  Dloss:0.4072  Gloss:9.2610 
ms per step:110.05  epoch:90/100  step:22/25  Dloss:0.3754  Gloss:9.0940 
ms per step:109.43  epoch:90/100  step:24/25  Dloss:0.5651  Gloss:8.9833 
ms per step:109.68  epoch:91/100  step:0/25  Dloss:0.4573  Gloss:9.2881 
ms per step:110.47  epoch:91/100  step:2/25  Dloss:0.4180  Gloss:8.7367 
ms per step:109.96  epoch:91/100  step:4/25  Dloss:0.2625  Gloss:9.1983 
ms per step:109.52  epoch:91/100  step:6/25  Dloss:0.3542  Gloss:9.0335 
ms per step:108.10  epoch:91/100  step:8/25  Dloss:0.9353  Gloss:9.6542 
ms per step:115.75  epoch:91/100  step:10/25  Dloss:0.5399  Gloss:9.5484 
ms per step:121.84  epoch:91/100  step:12/25  Dloss:0.3733  Gloss:9.5674 
ms per step:114.77  epoch:91/100  step:14/25  Dloss:0.3088  Gloss:9.4505 
ms per step:111.03  epoch:91/100  step:16/25  Dloss:0.2578  Gloss:10.2645 
ms per step:108.86  epoch:91/100  step:18/25  Dloss:0.4453  Gloss:9.1961 
ms per step:108.47  epoch:91/100  step:20/25  Dloss:0.6168  Gloss:9.7312 
ms per step:109.57  epoch:91/100  step:22/25  Dloss:0.5744  Gloss:9.5293 
ms per step:108.97  epoch:91/100  step:24/25  Dloss:0.3736  Gloss:9.7246 
ms per step:109.59  epoch:92/100  step:0/25  Dloss:0.3723  Gloss:8.5488 
ms per step:111.46  epoch:92/100  step:2/25  Dloss:0.2898  Gloss:8.7300 
ms per step:109.57  epoch:92/100  step:4/25  Dloss:0.3565  Gloss:9.6917 
ms per step:112.70  epoch:92/100  step:6/25  Dloss:0.3713  Gloss:9.4799 
ms per step:109.44  epoch:92/100  step:8/25  Dloss:0.5026  Gloss:9.7731 
ms per step:130.81  epoch:92/100  step:10/25  Dloss:0.3742  Gloss:11.1001 
ms per step:113.76  epoch:92/100  step:12/25  Dloss:0.3412  Gloss:9.7544 
ms per step:113.85  epoch:92/100  step:14/25  Dloss:0.4109  Gloss:8.3546 
ms per step:110.97  epoch:92/100  step:16/25  Dloss:0.4218  Gloss:8.6867 
ms per step:109.30  epoch:92/100  step:18/25  Dloss:0.5266  Gloss:9.3781 
ms per step:107.89  epoch:92/100  step:20/25  Dloss:0.4983  Gloss:9.4584 
ms per step:107.67  epoch:92/100  step:22/25  Dloss:0.4492  Gloss:10.1151 
ms per step:108.39  epoch:92/100  step:24/25  Dloss:0.3090  Gloss:8.9276 
ms per step:109.53  epoch:93/100  step:0/25  Dloss:0.4238  Gloss:8.6132 
ms per step:108.22  epoch:93/100  step:2/25  Dloss:0.5983  Gloss:9.5171 
ms per step:109.87  epoch:93/100  step:4/25  Dloss:0.4518  Gloss:9.2334 
ms per step:108.43  epoch:93/100  step:6/25  Dloss:0.5205  Gloss:8.6449 
ms per step:108.50  epoch:93/100  step:8/25  Dloss:0.4218  Gloss:9.0550 
ms per step:113.74  epoch:93/100  step:10/25  Dloss:0.3159  Gloss:9.4041 
ms per step:112.95  epoch:93/100  step:12/25  Dloss:0.3870  Gloss:10.0409 
ms per step:114.25  epoch:93/100  step:14/25  Dloss:0.5505  Gloss:10.1617 
ms per step:113.36  epoch:93/100  step:16/25  Dloss:0.5675  Gloss:9.0700 
ms per step:109.72  epoch:93/100  step:18/25  Dloss:0.4776  Gloss:9.0256 
ms per step:107.72  epoch:93/100  step:20/25  Dloss:0.4682  Gloss:9.4622 
ms per step:109.42  epoch:93/100  step:22/25  Dloss:0.3775  Gloss:9.3905 
ms per step:111.35  epoch:93/100  step:24/25  Dloss:0.4195  Gloss:8.7059 
ms per step:110.19  epoch:94/100  step:0/25  Dloss:0.3528  Gloss:10.4851 
ms per step:110.85  epoch:94/100  step:2/25  Dloss:0.3705  Gloss:9.7367 
ms per step:112.05  epoch:94/100  step:4/25  Dloss:0.3281  Gloss:9.7149 
ms per step:108.98  epoch:94/100  step:6/25  Dloss:0.4161  Gloss:9.2867 
ms per step:108.63  epoch:94/100  step:8/25  Dloss:0.4961  Gloss:9.0830 
ms per step:113.79  epoch:94/100  step:10/25  Dloss:0.3964  Gloss:9.4060 
ms per step:113.84  epoch:94/100  step:12/25  Dloss:0.5112  Gloss:9.2747 
ms per step:114.72  epoch:94/100  step:14/25  Dloss:0.4263  Gloss:9.8867 
ms per step:113.31  epoch:94/100  step:16/25  Dloss:0.4281  Gloss:9.9811 
ms per step:109.11  epoch:94/100  step:18/25  Dloss:0.4899  Gloss:8.9997 
ms per step:108.89  epoch:94/100  step:20/25  Dloss:0.4239  Gloss:8.1569 
ms per step:109.12  epoch:94/100  step:22/25  Dloss:0.6578  Gloss:9.2922 
ms per step:108.82  epoch:94/100  step:24/25  Dloss:0.4397  Gloss:8.8495 
ms per step:108.65  epoch:95/100  step:0/25  Dloss:0.3590  Gloss:9.7673 
ms per step:108.55  epoch:95/100  step:2/25  Dloss:0.3656  Gloss:9.7763 
ms per step:108.13  epoch:95/100  step:4/25  Dloss:0.3840  Gloss:9.6572 
ms per step:108.67  epoch:95/100  step:6/25  Dloss:0.3910  Gloss:10.0574 
ms per step:114.06  epoch:95/100  step:8/25  Dloss:0.5389  Gloss:9.5577 
ms per step:122.28  epoch:95/100  step:10/25  Dloss:0.4720  Gloss:8.2654 
ms per step:121.15  epoch:95/100  step:12/25  Dloss:0.4213  Gloss:9.6620 
ms per step:121.15  epoch:95/100  step:14/25  Dloss:0.4435  Gloss:8.9988 
ms per step:119.30  epoch:95/100  step:16/25  Dloss:0.4864  Gloss:9.3086 
ms per step:115.39  epoch:95/100  step:18/25  Dloss:0.5011  Gloss:8.6628 
ms per step:115.83  epoch:95/100  step:20/25  Dloss:0.5114  Gloss:9.4756 
ms per step:118.84  epoch:95/100  step:22/25  Dloss:0.4530  Gloss:8.8176 
ms per step:118.08  epoch:95/100  step:24/25  Dloss:0.4862  Gloss:9.1391 
ms per step:118.40  epoch:96/100  step:0/25  Dloss:0.4929  Gloss:9.3366 
ms per step:117.15  epoch:96/100  step:2/25  Dloss:0.5273  Gloss:9.7053 
ms per step:117.04  epoch:96/100  step:4/25  Dloss:0.3610  Gloss:10.0717 
ms per step:116.71  epoch:96/100  step:6/25  Dloss:0.3710  Gloss:9.0325 
ms per step:115.58  epoch:96/100  step:8/25  Dloss:0.4207  Gloss:9.5342 
ms per step:119.91  epoch:96/100  step:10/25  Dloss:0.7068  Gloss:8.9157 
ms per step:120.43  epoch:96/100  step:12/25  Dloss:0.7362  Gloss:9.0200 
ms per step:120.88  epoch:96/100  step:14/25  Dloss:0.6148  Gloss:9.1065 
ms per step:120.18  epoch:96/100  step:16/25  Dloss:0.4713  Gloss:9.4374 
ms per step:116.16  epoch:96/100  step:18/25  Dloss:0.4350  Gloss:9.1553 
ms per step:115.53  epoch:96/100  step:20/25  Dloss:0.3483  Gloss:9.1159 
ms per step:117.04  epoch:96/100  step:22/25  Dloss:0.4225  Gloss:9.3998 
ms per step:116.63  epoch:96/100  step:24/25  Dloss:0.3894  Gloss:8.8221 
ms per step:115.70  epoch:97/100  step:0/25  Dloss:0.2969  Gloss:11.4602 
ms per step:116.13  epoch:97/100  step:2/25  Dloss:0.3374  Gloss:8.8577 
ms per step:116.38  epoch:97/100  step:4/25  Dloss:0.4653  Gloss:10.1228 
ms per step:113.70  epoch:97/100  step:6/25  Dloss:0.5599  Gloss:8.7278 
ms per step:115.82  epoch:97/100  step:8/25  Dloss:0.4568  Gloss:9.1296 
ms per step:121.05  epoch:97/100  step:10/25  Dloss:0.3903  Gloss:9.8830 
ms per step:120.40  epoch:97/100  step:12/25  Dloss:0.4433  Gloss:10.3950 
ms per step:114.96  epoch:97/100  step:14/25  Dloss:0.3300  Gloss:9.2907 
ms per step:114.69  epoch:97/100  step:16/25  Dloss:0.4651  Gloss:8.2343 
ms per step:118.24  epoch:97/100  step:18/25  Dloss:0.5483  Gloss:9.1674 
ms per step:118.30  epoch:97/100  step:20/25  Dloss:0.5074  Gloss:8.7715 
ms per step:116.01  epoch:97/100  step:22/25  Dloss:0.4239  Gloss:8.6857 
ms per step:114.81  epoch:97/100  step:24/25  Dloss:0.3656  Gloss:9.4334 
ms per step:113.52  epoch:98/100  step:0/25  Dloss:0.5340  Gloss:8.7831 
ms per step:114.07  epoch:98/100  step:2/25  Dloss:0.5204  Gloss:9.3759 
ms per step:115.33  epoch:98/100  step:4/25  Dloss:0.4783  Gloss:10.1390 
ms per step:115.68  epoch:98/100  step:6/25  Dloss:0.5141  Gloss:9.6782 
ms per step:115.75  epoch:98/100  step:8/25  Dloss:0.4573  Gloss:10.2644 
ms per step:122.51  epoch:98/100  step:10/25  Dloss:0.3828  Gloss:9.3207 
ms per step:118.74  epoch:98/100  step:12/25  Dloss:0.3671  Gloss:9.4600 
ms per step:122.06  epoch:98/100  step:14/25  Dloss:0.4081  Gloss:8.8109 
ms per step:112.10  epoch:98/100  step:16/25  Dloss:0.3973  Gloss:9.4212 
ms per step:110.98  epoch:98/100  step:18/25  Dloss:0.4818  Gloss:9.2266 
ms per step:108.19  epoch:98/100  step:20/25  Dloss:0.5731  Gloss:8.4570 
ms per step:106.78  epoch:98/100  step:22/25  Dloss:0.5965  Gloss:9.1554 
ms per step:107.09  epoch:98/100  step:24/25  Dloss:0.6044  Gloss:9.3033 
ms per step:107.77  epoch:99/100  step:0/25  Dloss:0.6031  Gloss:9.0418 
ms per step:107.35  epoch:99/100  step:2/25  Dloss:0.4760  Gloss:8.5322 
ms per step:107.44  epoch:99/100  step:4/25  Dloss:0.3417  Gloss:8.8869 
ms per step:108.21  epoch:99/100  step:6/25  Dloss:0.4138  Gloss:8.9011 
ms per step:107.55  epoch:99/100  step:8/25  Dloss:0.3341  Gloss:9.8052 
ms per step:112.92  epoch:99/100  step:10/25  Dloss:0.4165  Gloss:9.5205 
ms per step:113.04  epoch:99/100  step:12/25  Dloss:0.4862  Gloss:8.9653 
ms per step:115.19  epoch:99/100  step:14/25  Dloss:0.5382  Gloss:9.0894 
ms per step:113.17  epoch:99/100  step:16/25  Dloss:0.4197  Gloss:9.1876 
ms per step:110.12  epoch:99/100  step:18/25  Dloss:0.5220  Gloss:9.8974 
ms per step:109.67  epoch:99/100  step:20/25  Dloss:0.3995  Gloss:9.1445 
ms per step:115.75  epoch:99/100  step:22/25  Dloss:0.4541  Gloss:8.3840 
ms per step:115.30  epoch:99/100  step:24/25  Dloss:0.4710  Gloss:9.3341 
ms per step:115.22  epoch:100/100  step:0/25  Dloss:0.4944  Gloss:9.2594 
ms per step:115.19  epoch:100/100  step:2/25  Dloss:0.5674  Gloss:9.4491 
ms per step:116.61  epoch:100/100  step:4/25  Dloss:0.6164  Gloss:9.5221 
ms per step:116.62  epoch:100/100  step:6/25  Dloss:0.4499  Gloss:9.0963 
ms per step:114.83  epoch:100/100  step:8/25  Dloss:0.3913  Gloss:9.3877 
ms per step:115.59  epoch:100/100  step:10/25  Dloss:0.4446  Gloss:9.7656 
ms per step:115.14  epoch:100/100  step:12/25  Dloss:0.5224  Gloss:8.4429 
ms per step:116.18  epoch:100/100  step:14/25  Dloss:0.5398  Gloss:8.8629 
ms per step:115.41  epoch:100/100  step:16/25  Dloss:0.6848  Gloss:8.9019 
ms per step:116.34  epoch:100/100  step:18/25  Dloss:0.5459  Gloss:9.1481 
ms per step:114.94  epoch:100/100  step:20/25  Dloss:0.5987  Gloss:9.4928 
ms per step:116.42  epoch:100/100  step:22/25  Dloss:0.5061  Gloss:8.3426 
ms per step:114.89  epoch:100/100  step:24/25  Dloss:0.4972  Gloss:8.8630 
CPU times: user 20min 45s, sys: 5min 35s, total: 26min 20s
Wall time: 4min 58s

推理

获取上述训练过程完成后的ckpt文件,通过load_checkpoint和load_param_into_net将ckpt中的权重参数导入到模型中,获取数据进行推理并对推理的效果图进行演示(由于时间问题,训练过程只进行了3个epoch,可根据需求调整epoch)。

from mindspore import load_checkpoint, load_param_into_netparam_g = load_checkpoint(ckpt_dir + "Generator.ckpt")
load_param_into_net(net_generator, param_g)
dataset = ds.MindDataset("./dataset/dataset_pix2pix/train.mindrecord", columns_list=["input_images", "target_images"], shuffle=True)
data_iter = next(dataset.create_dict_iterator())
predict_show = net_generator(data_iter["input_images"])
plt.figure(figsize=(10, 3), dpi=140)
for i in range(10):plt.subplot(2, 10, i + 1)plt.imshow((data_iter["input_images"][i].asnumpy().transpose(1, 2, 0) + 1) / 2)plt.axis("off")plt.subplots_adjust(wspace=0.05, hspace=0.02)plt.subplot(2, 10, i + 11)plt.imshow((predict_show[i].asnumpy().transpose(1, 2, 0) + 1) / 2)plt.axis("off")plt.subplots_adjust(wspace=0.05, hspace=0.02)
plt.show()

各数据集分别推理的效果如下

pix2pix3

引用

[1] Phillip Isola,Jun-Yan Zhu,Tinghui Zhou,Alexei A. Efros. Image-to-Image Translation with Conditional Adversarial Networks.[J]. CoRR,2016,abs/1611.07004.

本文来自互联网用户投稿,该文观点仅代表作者本人,不代表本站立场。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如若转载,请注明出处:http://www.rhkb.cn/news/438301.html

如若内容造成侵权/违法违规/事实不符,请联系长河编程网进行投诉反馈email:809451989@qq.com,一经查实,立即删除!

相关文章

HUAWEI New4.9G 与 2.6G 无法正常切换问题处理案例

HUAWEI New4.9G 与 2.6G 无法正常切换问题处理案例 在某地市的 XX 音乐节保障准备期间&#xff0c;为确保活动期间的网络质量&#xff0c;现场新开了 4.9G HUAWEI 室外基站。在网络优化和测试中&#xff0c;发现UE无法实现从 2.6G 到 4.9G 的正常切换。虽然现场具备 4.9G信号覆…

Python | Leetcode Python题解之第448题找到所有数组中消失的数字

题目&#xff1a; 题解&#xff1a; class Solution:def findDisappearedNumbers(self, nums: List[int]) -> List[int]:n len(nums)for num in nums:x (num - 1) % nnums[x] nret [i 1 for i, num in enumerate(nums) if num < n]return ret

YOLOv8 结合设计硬件感知神经网络设计的高效 Repvgg的ConvNet 网络结构 ,改进EfficientRep结构

一、理论部分 摘要—我们提出了一种硬件高效的卷积神经网络架构,它具有类似 repvgg 的架构。Flops 或参数是评估网络效率的传统指标,这些网络对硬件(包括计算能力和内存带宽)不敏感。因此,如何设计神经网络以有效利用硬件的计算能力和内存带宽是一个关键问题。本文提出了一…

1、Spring Boot 3.x 集成 Eureka Server/Client

一、前言 基于 Spring Boot 3.x 版本开发&#xff0c;因为 Spring Boot 3.x 暂时没有正式发布&#xff0c;所以很少有 Spring Boot 3.x 开发的项目&#xff0c;自己也很想了踩踩坑&#xff0c;看看 Spring Boot 3.x 与 2.x 有什么区别。自己与记录一下在 Spring Boot 3.x 过程…

exe4j安装使用教程

A-XVK258563F-1p4lv7mg7sav A-XVK209982F-1y0i3h4ywx2h1 A-XVK267351F-dpurrhnyarva A-XVK204432F-1kkoilo1jy2h3r A-XVK246130F-1l7msieqiwqnq A-XVK249554F-pllh351kcke50

第5篇:MySQL日志分析----应急响应之日志分析篇

常见的数据库攻击包括弱口令、SQL注入、提升权限、窃取备份等。对数据库日志进行分析&#xff0c;可以发现攻击行为&#xff0c;进一步还原攻击场景及追溯攻击源。 0x01 Mysql日志分析 general query log能记录成功连接和每次执行的查询&#xff0c;我们可以将它用作安全布防…

Android SystemUI组件(08)睡眠灭屏 锁屏处理流程

该系列文章总纲链接&#xff1a;专题分纲目录 Android SystemUI组件 本章关键点总结 & 说明&#xff1a; 说明&#xff1a;本章节持续迭代之前章节的思维导图&#xff0c;主要关注左侧上方锁屏分析部分 睡眠灭屏 即可。 Power按键的处理逻辑最终是由PhoneWindowManager来完…

【数据结构】图的最小生成树

快乐的流畅&#xff1a;个人主页 个人专栏&#xff1a;《C游记》《进击的C》《Linux迷航》 远方有一堆篝火&#xff0c;在为久候之人燃烧&#xff01; 文章目录 引言一、最小生成树的概念二、Kruskal算法2.1 思想2.2 实现 三、Prim算法3.1 思想3.2 实现 四、Kruskal和Prim的对比…

Spring Task 调度任务

Spring Task是调度任务框架&#xff0c;通过配置&#xff0c;程序可以按照约定的时间自动执行代码逻辑&#xff0c;基于注解方式实现需要如下注解&#xff1a; Component 任务调度类交给Spring IOC容器管理EnableScheduling 启用 Spring 的定时任务&#xff08;Scheduling&…

索尼MDR-M1:超宽频的音频盛宴,打造沉浸式音乐体验

在音乐的世界里&#xff0c;每一次技术的突破都意味着全新的听觉体验。 索尼&#xff0c;作为音频技术的先锋&#xff0c;再次以其最新力作——MDR-M1封闭式监听耳机&#xff0c;引领了音乐界的新潮流。 这款耳机以其超宽频播放和卓越的隔音性能&#xff0c;为音乐爱好者和专…

k8s中,ingress的实现原理,及其架构。

图片来源&#xff1a;自己画的 图片来源&#xff1a;k8s官网 首先&#xff0c;什么是ingress? 是服务还是控制器&#xff1f; 都不精确 ingress是一个api资源 service和deployment也是api资源。 这几个相互协作&#xff0c;组建成一个对外提供服务的架构。 ingress提供的…

[C++]使用纯opencv部署yolov11目标检测onnx模型

yolov11官方框架&#xff1a;https://github.com/ultralytics/ultralytics 【算法介绍】 在C中使用纯OpenCV部署YOLOv11进行目标检测是一项具有挑战性的任务&#xff0c;因为YOLOv11通常是用PyTorch等深度学习框架实现的&#xff0c;而OpenCV本身并不直接支持加载和运行PyTor…

系统安全 - RedisMySQL安全及实践

文章目录 导图Redis 安全潜在的安全风险防护措施密码认证命令重命名权限最小化日志和审计 Red网络隔离 MySQL 安全认证和授权文件操作风险传输和存储加密最小权限原则审计 总结 导图 Redis 安全 Redis的设计初衷是为了在可信环境下提供高性能的KV数据库服务&#xff0c;因此它…

FiBiNET模型实现推荐算法

1. 项目简介 A031-FiBiNET模型项目是一个基于深度学习的推荐系统算法实现&#xff0c;旨在提升推荐系统的性能和精度。该项目的背景源于当今互联网平台中&#xff0c;推荐算法在电商、社交、内容分发等领域的广泛应用。推荐系统通过分析用户的历史行为和兴趣偏好&#xff0c;预…

【NIO基础】NIO(非阻塞 I/O)和 IO(传统 I/O)的区别,以及 NIO 的三大组件详解

目录 1、NIO 2、NIO 和 IO 的区别 1. 阻塞 vs 非阻塞 2. 一个线程 vs 多个连接 3. 面向流 vs 面向缓冲 4. 多路复用 3、Channel & Buffer (1&#xff09;Channel&#xff1a;双向通道 (2&#xff09;Buffer&#xff1a;缓冲区 (3&#xff09;ByteBuffer&#xff…

用Arduino单片机读取PCF8591模数转换器的模拟量并转化为数字输出

PCF8591是一款单芯片&#xff0c;单电源和低功耗8位CMOS数据采集设备。博文[1]对该产品已有介绍&#xff0c;此处不再赘述。但该博文是使用NVIDIA Jetson nano运行python读取输入PCF8591的模拟量的&#xff0c;读取的结果显示在屏幕上&#xff0c;或输出模拟量点亮灯。NVIDIA J…

计算机毕业设计 基于Python的智能文献管理系统的设计与实现 Python+Django+Vue 前后端分离 附源码 讲解 文档

&#x1f34a;作者&#xff1a;计算机编程-吉哥 &#x1f34a;简介&#xff1a;专业从事JavaWeb程序开发&#xff0c;微信小程序开发&#xff0c;定制化项目、 源码、代码讲解、文档撰写、ppt制作。做自己喜欢的事&#xff0c;生活就是快乐的。 &#x1f34a;心愿&#xff1a;点…

讯飞星火编排创建智能体学习(四):网页读取

目录 引言 网页读取节点 如何生成网址 测试 引言 在讯飞星火编排创建智能体学习&#xff08;三&#xff09;&#xff1a;搜索工具-CSDN博客中&#xff0c;我介绍了如何用搜索工具从网上搜索车次信息。不过&#xff0c;在测试中我们也发现讯飞星火的这个工具并不是特别完善&…

誉天Linux云计算课程学什么?为什么保障就业?

一个IT工程师相当于干了哪些职业? 其中置顶回答生动而形象地描绘道&#xff1a; 一个IT工程师宛如一个超级多面手&#xff0c;相当于——加班狂程序员测试工程师实施工程师网络工程师电工装卸工搬运工超人。 此中酸甜苦辣咸&#xff0c;相信很多小伙伴们都深有体会。除了典…

ESP01 AT指令学习

一 、AT指令 测试指令&#xff1a;ATCWMODE? 参数及取值范围 cwmode&#xff08;1-3&#xff09; 查询指令&#xff1a; ATCWMODE&#xff1f; 当前cwmode的取值 3 设置指令&#xff1a; ATCWMODE3 设置当前的cwmode为 3 1、station 模式 连接到其他wifi 2、softA…