allegroai

在做试验时，需要画图，需要统计性能，查看试验的进度，如果全部都靠jupyter，多少有些繁琐，因此想要找一些pipeline工具，最好可以可视化查看，因此调研了allegroai，如下。

trains是开源的python组件，包括Trains Client, Trains Server, Trains agent

 一、Trains Server

如果不指定，默认会将结果上传到官方的demo网站，因此需要自己安装配置Server
### 安装
需要安装docker-compose
git 路径： https://github.com/allegroai/trains-server

安装需要注意的是：
1. chmod -R 777   /opt/trains/
2. 需要将docker-compose.yml放到/opt/trains/

配置

server会有一些配置
https://allegro.ai/docs/deploying_trains/trains_server_config/
默认的配置文件 trains-server/server/config/default/
例如，对登录网站的用户名和密码设置/opt/trains/config/apiserver.conf

设置完之后重启

docker-compose down
docker-compose -f docker-compose.yml up -d

 二、Trains Python Client

A. 配置使用某台服务器搭建的Server

不使用默认的demo wep ui，而是自己配置的server，以下2种方法
1. trains-init
然后按照步骤去操作
2.   ~/trains.conf
 

api {# API server on port 8008api_server: "http://localhost:8008"# web_server on port 8080web_server: "http://localhost:8080"# file server on port 8081files_server: "http://localhost:8081"}

B. 安装

server提供了可视化的Web UI,我们只需在自己的脚本中添加几行代码，即可。

pip install trains

C. 使用例子

函数说明文档API：https://allegro.ai/docs/logger.html

举例说明，以下面脚本为例
https://github.com/allegroai/trains/blob/master/examples/frameworks/pytorch/pytorch_mnist.py

from trains import Task
task = Task.init(project_name="my project", task_name="my task")

1. output 指定输出路径
Task.init()
参数 output_uri = model_snapshots_path
```
task = Task.init(project_name='pkl examples', task_name='pkl train',output_uri='/allegro/trians/')
```
脚本运行之后，会创建以下路径
 

+--

|   +--
|       +-- .
|           +-- models
|           +-- artifacts

2. logger 日志：文字或画图
可视化绘图

logger = task.get_logger()
logger.report_histogram()
logger.report_text()

有时候想要将dataframe也上传到plot上，会更好看一点

from trains import Task
task = Task.init(project_name='Training', task_name='pandas table reporting')logger = task.get_logger()logger.report_table("table pd", "table pd",iteration=0, table_plot=df)task.close()

结果会在result标签下的plots子标签

3. register artifact 会上传到server，change也会上传
artifact可以认为是将dataframe上传到server查看

# Register the test loss and correct as a Pandas DataFrame artifact
task.register_artifact('Test_Loss_Correct', df, metadata={'metadata string': 'apple', 'metadata int': 100, 'metadata dict': {'dict string': 'pear', 'dict int': 200}})
# Once the artifact is registered, we can get it and work with it. Here, we sample it.
sample = Task.current_task().get_registered_artifacts()['Test_Loss_Correct'].sample(frac=0.5, replace=True, random_state=1)

4. task.upload_artifact()  会上传到server，但是change不会

三、实例

以下是个完整的例子和结果

# TRAINS - Example of Pytorch mnist training integration
#
from __future__ import print_function
import argparse
import os
from tempfile import gettempdirimport torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import numpy as np
import pandas as pd
from torchvision import datasets, transformsfrom trains import Task, Loggerclass Net(nn.Module):def __init__(self):super(Net, self).__init__()self.conv1 = nn.Conv2d(1, 20, 5, 1)self.conv2 = nn.Conv2d(20, 50, 5, 1)self.fc1 = nn.Linear(4 * 4 * 50, 500)self.fc2 = nn.Linear(500, 10)def forward(self, x):x = F.relu(self.conv1(x))x = F.max_pool2d(x, 2, 2)x = F.relu(self.conv2(x))x = F.max_pool2d(x, 2, 2)x = x.view(-1, 4 * 4 * 50)x = F.relu(self.fc1(x))x = self.fc2(x)return F.log_softmax(x, dim=1)def train(args, model, device, train_loader, optimizer, epoch):model.train()for batch_idx, (data, target) in enumerate(train_loader):data, target = data.to(device), target.to(device)optimizer.zero_grad()output = model(data)loss = F.nll_loss(output, target)loss.backward()optimizer.step()if batch_idx % args.log_interval == 0:####以下是scalarLogger.current_logger().report_scalar("train", "loss", iteration=(epoch * len(train_loader) + batch_idx), value=loss.item())print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(epoch, batch_idx * len(data), len(train_loader.dataset),100. * batch_idx / len(train_loader), loss.item()))# Add manual scalar reporting for loss metricsLogger.current_logger().report_scalar(title='Scalar example {} - epoch'.format(epoch), series='Loss', value=loss.item(), iteration=batch_idx)def test(args, model, device, test_loader, epoch):model.eval()test_loss = 0correct = 0save_test_loss = []save_correct = []with torch.no_grad():for data, target in test_loader:data, target = data.to(device), target.to(device)output = model(data)test_loss += F.nll_loss(output, target, reduction='sum').item()  # sum up batch losspred = output.argmax(dim=1, keepdim=True)  # get the index of the max log-probabilitycorrect += pred.eq(target.view_as(pred)).sum().item()save_correct.append(correct)save_test_loss.append(test_loss)test_loss /= len(test_loader.dataset)#####以下是scalarLogger.current_logger().report_scalar("test", "loss", iteration=epoch, value=test_loss)Logger.current_logger().report_scalar("test", "accuracy", iteration=epoch, value=(correct / len(test_loader.dataset)))print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(test_loss, correct, len(test_loader.dataset),100. * correct / len(test_loader.dataset)))###以下是画图Logger.current_logger().report_histogram(title='Histogram example', series='correct',iteration=1, values=save_correct, xaxis='Test', yaxis='Correct')# Manually report test loss and correct as a confusion matrixmatrix = np.array([save_test_loss, save_correct])Logger.current_logger().report_confusion_matrix(title='Confusion matrix example', series='Test loss / correct', matrix=matrix, iteration=1)# Create the Pandas DataFrametest_loss_correct = {'test lost': save_test_loss,'correct': save_correct}df = pd.DataFrame(test_loss_correct, columns=['test lost','correct'])##可以将dataframe上传到server查看# Register the test loss and correct as a Pandas DataFrame artifactTask.current_task().register_artifact('Test_Loss_Correct', df, metadata={'metadata string': 'apple', 'metadata int': 100, 'metadata dict': {'dict string': 'pear', 'dict int': 200}})# Upload test loss as an artifact. Here, the artifact is numpy arrayTask.current_task().upload_artifact('Predictions',artifact_object=np.array(save_test_loss),metadata={'metadata string': 'banana', 'metadata integer': 300,'metadata dictionary': {'dict string': 'orange', 'dict int': 400}})def main():task = Task.init(project_name='pkl examples', task_name='pkl train',output_uri='/allegro/trians/')# Training settingsparser = argparse.ArgumentParser(description='PyTorch MNIST Example')parser.add_argument('--batch-size', type=int, default=64, metavar='N',help='input batch size for training (default: 64)')parser.add_argument('--test-batch-size', type=int, default=1000, metavar='N',help='input batch size for testing (default: 1000)')parser.add_argument('--epochs', type=int, default=10, metavar='N',help='number of epochs to train (default: 10)')parser.add_argument('--lr', type=float, default=0.01, metavar='LR',help='learning rate (default: 0.01)')parser.add_argument('--momentum', type=float, default=0.5, metavar='M',help='SGD momentum (default: 0.5)')parser.add_argument('--no-cuda', action='store_true', default=False,help='disables CUDA training')parser.add_argument('--seed', type=int, default=1, metavar='S',help='random seed (default: 1)')parser.add_argument('--log-interval', type=int, default=10, metavar='N',help='how many batches to wait before logging training status')parser.add_argument('--save-model', action='store_true', default=True,help='For Saving the current Model')args = parser.parse_args()use_cuda = not args.no_cuda and torch.cuda.is_available()torch.manual_seed(args.seed)device = torch.device("cuda" if use_cuda else "cpu")kwargs = {'num_workers': 4, 'pin_memory': True} if use_cuda else {}train_loader = torch.utils.data.DataLoader(datasets.MNIST(os.path.join('..', 'data'), train=True, download=True,transform=transforms.Compose([transforms.ToTensor(),transforms.Normalize((0.1307,), (0.3081,))])),batch_size=args.batch_size, shuffle=True, **kwargs)test_loader = torch.utils.data.DataLoader(datasets.MNIST(os.path.join('..', 'data'), train=False, transform=transforms.Compose([transforms.ToTensor(),transforms.Normalize((0.1307,), (0.3081,))])),batch_size=args.test_batch_size, shuffle=True, **kwargs)model = Net().to(device)optimizer = optim.SGD(model.parameters(), lr=args.lr, momentum=args.momentum)for epoch in range(1, args.epochs + 1):train(args, model, device, train_loader, optimizer, epoch)test(args, model, device, test_loader, epoch)if (args.save_model):torch.save(model.state_dict(), os.path.join(gettempdir(), "mnist_cnn.pt"))if __name__ == '__main__':main()

可以通过网址： http://IP:8080/  查看脚本运行的进度，以及log等

查看 artifacts

查看log

查看scalars

查看plots，可交互图

本文来自互联网用户投稿，文章观点仅代表作者本人，不代表本站立场，不承担相关法律责任。如若转载，请注明出处。 如若内容造成侵权/违法违规/事实不符，请点击【内容举报】进行投诉反馈！