本文已参与「新人创造礼」活动，一起敞开创造之路。

本文首发于CSDN。

诸神缄默不语-个人CSDN博文目录

wandb官网：Weights & Biases – Developer tools for ML

最近更新时刻：2022.9.12 最早更新时刻：2022.5.17

账号注册和登录应该很简单，我就不说了。

@[toc]

1. 系统全体设置

在wandb.ai/settings中能够设置什么时候以邮件或slack（我没下，所以我全用邮件了）提示：

其中run finished在使用jupyter notebook环境时不会起效，以防每个cell运转完毕后都提示一次。假如用jupyter notebook的话，需求用wandb.alert()来触发提示。

此外还能够关联GitHub以迅速提交benchmark runs（这个我还没试过，所以还没有关联）：

贮存空间能够进行管理，免费的有100个G，点击“manage storage”能够直接删不需求的文件：

2. 学术

2.1 免费学术小组

这个我还没试过，意思应该是校园或科研机构之类的能够免费建小组。假如有需求能够使用：wandb.ai/site/academ…

2.2 在论文中引用wandb的方法

来自wandb.ai/site/academ…：

@misc{wandb,
title = {Experiment Tracking with Weights and Biases},
year = {2020},
note = {Software available from wandb.com},
url={https://www.wandb.com/},
author = {Biewald, Lukas},
}

2.3 wandb白皮书

www.dropbox.com/s/0ipub9eww…

3. wandb文档

Weights & Biases – Documentation

wandb支持多渠道、多框架。

3.1 wandb quickstart

Quickstart – Documentation Jupyter Notebook代码实践：Intro_to_Weights_&_Biases.ipynb – Colaboratory

3.1.1 set up

指令行：

pip install wandb
wandb login

（假如代码中有用到wandb，其实会主动要求登录，可是假如用nohup挂起脚本的话就无法完成，所以仍是建议提前登录好） API key的方位在：wandb.ai/authorize （值得注意的是，这个配置应该是大局的，尽管我在一个虚拟环境下登录了我的账号，可是在别的虚拟环境下重新安装wandb仍是能够直接通用。我估计是因为缓存在了本地） 用wandb login --relogin强制重新登录。

用jupyter notebook则为：

!pip install wandb
wandb.login()

3.1.2 敞开新项目，盯梢目标、超参，增加报警信息

感觉就是init新项目，然后目标用log（在下面比如里能够看到，能够分组（如分为train/val）），超参用config，需求报警的内容就增加alert，成果增加到summary，最终finish就行，比较好用。

init文档：Launch Experiments with wandb.init – Documentation 对dashboard全体布局（run page）的介绍：Run Page – Documentation 数据可视化/盯梢目标：Data Visualization – Documentation 盯梢超参：Configure Experiments with wandb.config – Documentation 正告部分文档：Send Alerts with wandb.alert – Documentation

在测验代码时如临时不想与wandb同步，需设置环境变量，使wandb模式变成离线： WANDB_MODE=offline（具体做法是在运转Python代码的指令行中，在python前面加上这句指令）

文档中给出的核心代码：

import wandb
wandb.init(project="my-awesome-project")  #其他入参：name（见下例），config（见下）
wandb.log({'accuracy': train_acc, 'loss': train_loss})
wandb.config.dropout = 0.2
wandb.alert(
    title="Low accuracy", 
    text=f"Accuracy {acc} is below the acceptable threshold {thresh}"
)

在jupyter notebook上，用一个假示例来模仿：

import random
# Launch 5 simulated experiments
total_runs = 5
for run in range(total_runs):
  #  1️⃣ Start a new run to track this script
  wandb.init(
      # Set the project where this run will be logged
      project="wandbexample1", 
      # We pass a run name (otherwise it’ll be randomly assigned, like sunshine-lollypop-10)
      name=f"experiment_{run}", 
      # Track hyperparameters and run metadata
      config={
      "learning_rate": 0.02,
      "architecture": "CNN",
      "dataset": "CIFAR-100",
      "epochs": 10,
      })
  # This simple block simulates a training loop logging metrics
  epochs = 10
  offset = random.random() / 5
  for epoch in range(2, epochs):
      acc = 1 - 2 ** -epoch - random.random() / epoch - offset
      loss = 2 ** -epoch + random.random() / epoch + offset
      #  2️⃣ Log metrics from your script to W&B
      wandb.log({"acc": acc, "loss": loss})
  # Mark the run as finished
  wandb.finish()

在输出中会给出wandb项目链接，在浏览器中打开即可。 记载的内容：

输出的内容：

能够看出运转数据保存在了本地的wandb文件夹，额外的信息都没有上传到wandb上。

3.1.3 上一节的PyTorch比如

jupyter notebook上：MNIST分类器

#@title
import wandb
import math
import random
import torch, torchvision
import torch.nn as nn
import torchvision.transforms as T
from tqdm.notebook import tqdm
device = "cuda:0" if torch.cuda.is_available() else "cpu"
def get_dataloader(is_train, batch_size, slice=5):
    "Get a training dataloader"
    full_dataset = torchvision.datasets.MNIST(root=".", train=is_train, transform=T.ToTensor(), download=True)
    sub_dataset = torch.utils.data.Subset(full_dataset, indices=range(0, len(full_dataset), slice))
    loader = torch.utils.data.DataLoader(dataset=sub_dataset, 
                                         batch_size=batch_size, 
                                         shuffle=True if is_train else False, 
                                         pin_memory=True, num_workers=2)
    return loader
def get_model(dropout):
    "A simple model"
    model = nn.Sequential(nn.Flatten(),
                         nn.Linear(28*28, 256),
                         nn.BatchNorm1d(256),
                         nn.ReLU(),
                         nn.Dropout(dropout),
                         nn.Linear(256,10)).to(device)
    return model
def validate_model(model, valid_dl, loss_func, log_images=False, batch_idx=0):
    "Compute performance of the model on the validation dataset and log a wandb.Table"
    model.eval()
    val_loss = 0.
    with torch.inference_mode():
        correct = 0
        for i, (images, labels) in tqdm(enumerate(valid_dl), leave=False):
            images, labels = images.to(device), labels.to(device)
            # Forward pass ➡
            outputs = model(images)
            val_loss += loss_func(outputs, labels)*labels.size(0)
            # Compute accuracy and accumulate
            _, predicted = torch.max(outputs.data, 1)
            correct += (predicted == labels).sum().item()
            # Log one batch of images to the dashboard, always same batch_idx.
            if i==batch_idx and log_images:
                log_image_table(images, predicted, labels, outputs.softmax(dim=1))
    return val_loss / len(valid_dl.dataset), correct / len(valid_dl.dataset)
def log_image_table(images, predicted, labels, probs):
    "Log a wandb.Table with (img, pred, target, scores)"
    #  Create a wandb Table to log images, labels and predictions to
    table = wandb.Table(columns=["image", "pred", "target"]+[f"score_{i}" for i in range(10)])
    for img, pred, targ, prob in zip(images.to("cpu"), predicted.to("cpu"), labels.to("cpu"), probs.to("cpu")):
        table.add_data(wandb.Image(img[0].numpy()*255), pred, targ, *prob.numpy())
    wandb.log({"predictions_table":table}, commit=False)

练习：

# Launch 5 experiments, trying different dropout rates
for i in range(5):
    #  initialise a wandb run
    wandb.init(
        project="wandbexample1",
        name="pytorch_example"+str(i),
        config={
            "epochs": 10,
            "batch_size": 128,
            "lr": 1e-3,
            "dropout": random.uniform(0.01, 0.80),
            })
    # Copy your config 
    config = wandb.config
    # Get the data
    train_dl = get_dataloader(is_train=True, batch_size=config.batch_size)
    valid_dl = get_dataloader(is_train=False, batch_size=2*config.batch_size)
    n_steps_per_epoch = math.ceil(len(train_dl.dataset) / config.batch_size)
    # A simple MLP model
    model = get_model(config.dropout)
    # Make the loss and optimizer
    loss_func = nn.CrossEntropyLoss()
    optimizer = torch.optim.Adam(model.parameters(), lr=config.lr)
   # Training
    example_ct = 0
    step_ct = 0
    for epoch in tqdm(range(config.epochs)):
        model.train()
        for step, (images, labels) in enumerate(tqdm(train_dl, leave=False)):
            images, labels = images.to(device), labels.to(device)
            outputs = model(images)
            train_loss = loss_func(outputs, labels)
            optimizer.zero_grad()
            train_loss.backward()
            optimizer.step()
            example_ct += len(images)
            metrics = {"train/train_loss": train_loss, 
                       "train/epoch": (step + 1 + (n_steps_per_epoch * epoch)) / n_steps_per_epoch, 
                       "train/example_ct": example_ct}
            if step + 1 < n_steps_per_epoch:
                #  Log train metrics to wandb 
                wandb.log(metrics)
            step_ct += 1
        val_loss, accuracy = validate_model(model, valid_dl, loss_func, log_images=(epoch==(config.epochs-1)))
        #  Log train and validation metrics to wandb
        val_metrics = {"val/val_loss": val_loss, 
                       "val/val_accuracy": accuracy}
        wandb.log({**metrics, **val_metrics})
        print(f"Train Loss: {train_loss:.3f}, Valid Loss: {val_loss:3f}, Accuracy: {accuracy:.2f}")
    # If you had a test set, this is how you could log it as a Summary metric
    wandb.summary['test_accuracy'] = 0.8
    #  Close your wandb run 
    wandb.finish()

wandb网页首页：

点进一个run内：

输出：

3.1.4 jupyter notebook wandb正告示例

# Start a wandb run
wandb.init(project="wandbexample1")
# Simulating a model training loop
acc_threshold = 0.3
for training_step in range(1000):
    # Generate a random number for accuracy
    accuracy = round(random.random() + random.random(), 3)
    print(f'Accuracy is: {accuracy}, {acc_threshold}')
    #  Log accuracy to wandb
    wandb.log({"Accuracy": accuracy})
    #  If the accuracy is below the threshold, fire a W&B Alert and stop the run
    if accuracy <= acc_threshold:
        #  Send the wandb Alert
        wandb.alert(
            title='Low Accuracy',
            text=f'Accuracy {accuracy} at step {training_step} is below the acceptable theshold, {acc_threshold}',
        )
        print('Alert triggered')
        break
# Mark the run as finished (useful in Jupyter notebooks)
wandb.finish()

acc变成0.155，小于阈值0.3，所以报了正告。因为我没有slack所以是给发到邮件上：

项目的dashboard：

3.2 调参

Hyperparameter Tuning – Documentation Organizing_Hyperparameter_Sweeps_in_PyTorch_with_W&B.ipynb

3.3 合作陈述

Collaborative Reports – Documentation

3.4 盯梢pipeline的数据和模型版别

Data + Model Versioning – Documentation

3.5 数据可视化/盯梢目标日志

Data Visualization – Documentation

3.6 主动化深度学习渠道的配置文件

Environment Variables – Documentation

3.7 本地化的解决方案

这个我暂时用不到，可是也列在这里以便备用： Private Hosting – Documentation

3.8 示例

Examples – Documentation 示例的dashboard：wandb_example Workspace – Weights & Biases

3.9 Interagations

Integrations – Documentation

3.10 输入输出贮存到wandb上的数据

Import & Export Data – Documentation

4. 常见问题

4.1 wandb.errors.UsageError: Error communicating with wandb process

参阅docs.wandb.ai/library/ini…：在wandb.init()中增加入参settings=wandb.Settings(start_method="fork")即可解决问题。

4.2 本地贮存空间不够→删去本地缓存

在 wandb.log() 处报错。

报错信息：

Thread SenderThread:
OSError: [Errno 28] No space left on device
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
  File "env_path/lib/python3.8/site-packages/wandb/sdk/internal/internal_util.py", line 51, in run
    self._run()
  File "env_path/lib/python3.8/site-packages/wandb/sdk/internal/internal_util.py", line 102, in _run
    self._process(record)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/internal/internal.py", line 310, in _process
    self._sm.send(record)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/internal/sender.py", line 304, in send
    send_handler(record)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/internal/sender.py", line 931, in send_summary
    self._update_summary()
  File "env_path/lib/python3.8/site-packages/wandb/sdk/internal/sender.py", line 944, in _update_summary
    f.write(json_summary)
OSError: [Errno 28] No space left on device
wandb: ERROR Internal wandb error: file data was not synced
Traceback (most recent call last):
  File "c49.py", line 293, in <module>
    wandb.log({'valid_law_acc':law_accuracy,'valid_charge_acc':charge_accuracy,
  File "env_path/lib/python3.8/site-packages/wandb/sdk/wandb_run.py", line 289, in wrapper
    return func(self, *args, **kwargs)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/wandb_run.py", line 255, in wrapper
    return func(self, *args, **kwargs)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/wandb_run.py", line 1591, in log
    self._log(data=data, step=step, commit=commit)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/wandb_run.py", line 1375, in _log
    self._partial_history_callback(data, step, commit)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/wandb_run.py", line 1259, in _partial_history_callback
    self._backend.interface.publish_partial_history(
  File "env_path/lib/python3.8/site-packages/wandb/sdk/interface/interface.py", line 553, in publish_partial_history
    self._publish_partial_history(partial_history)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/interface/interface_shared.py", line 67, in _publish_partial_history
    self._publish(rec)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/interface/interface_sock.py", line 51, in _publish
    self._sock_client.send_record_publish(record)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/lib/sock_client.py", line 150, in send_record_publish
    self.send_server_request(server_req)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/lib/sock_client.py", line 84, in send_server_request
    self._send_message(msg)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/lib/sock_client.py", line 81, in _send_message
    self._sendall_with_error_handle(header + data)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/lib/sock_client.py", line 61, in _sendall_with_error_handle
    sent = self._sock.send(data[total_sent:])
BrokenPipeError: [Errno 32] Broken pipe
Error in atexit._run_exitfuncs:
Traceback (most recent call last):
  File "env_path/lib/python3.8/site-packages/wandb/sdk/lib/sock_client.py", line 81, in _send_message
    self._sendall_with_error_handle(header + data)
  File "env_path/lib/python3.8/site-packages/wandb/sdk/lib/sock_client.py", line 61, in _sendall_with_error_handle
    sent = self._sock.send(data[total_sent:])
BrokenPipeError: [Errno 32] Broken pipe

解决方案：wandb sync --clean （参阅docs.wandb.ai/ref/cli/wan…）