Introduction to WanDB

A. What is WanDB?

Weights & Biases (WanDB) is a developer-oriented toolset, designed specifically for machine learning. It helps monitor and visualize the model’s training process and its performance in a more intuitive way. WanDB provides a centralized platform where teams can log, share, and collaborate on their machine learning projects, making it easier to compare different runs and models’ performance.

B. Why use WanDB?

There are several reasons why WanDB stands out as a preferred tool for machine learning projects:

  1. Track and Visualize Models: WanDB provides a simple way to track every detail of your experiment, providing real-time visualization of your models’ training and results.

  2. Hyperparameter Optimization: With WanDB’s Sweeps, you can automate hyperparameter tuning and explore the parameter space more efficiently to optimize your model’s performance.

  3. Collaboration: WanDB makes it easy to share your experiment results with colleagues and the community, fostering collaboration and knowledge sharing.

  4. Reproducibility: By logging all the metadata from your runs, WanDB helps maintain the reproducibility of your experiments, which is crucial in machine learning projects.

C. Understanding the Importance of Monitoring and Improving Model Training

Monitoring and improving model training is an essential part of the machine learning workflow.

  • Monitoring: By keeping track of various metrics such as loss and accuracy during the training process, you can understand how well your model is learning and whether it’s improving over time. This can help you detect issues like overfitting or underfitting early on and take corrective actions.

  • Improving: Once you monitor your model’s performance, the next step is to improve it. This could involve tweaking the model’s architecture, optimizing hyperparameters, or using more/ different data for training. Tools like WanDB make it easier to experiment with these aspects and track the impact of each change, thereby helping you build better models.

Setting up WanDB

A. Account Setup

To get started with WanDB, you need to create a free account:

  1. Visit the official Weights & Biases website: wandb.ai
  2. Click on Sign Up at the top right corner of the home page.
  3. You can opt to sign up using a GitHub, Google, or LinkedIn account. Alternatively, sign up using your email address and a password.
  4. You will see your API Keys/Token, you will need it. If you didn’t save it or you lost it, you can look it up in User Settings.

B. Installing the Wandb library

Once you’ve set up your account, you need to install the Wandb library in your Python environment. It can be installed using pip:

pip install wandb

Or with conda:

conda install -c conda-forge wandb

Ensure you have the latest version of the library for optimal functionality.

C. Initializing WanDB in Your Code

After installing the Wandb library, you need to import it and initialize it within your project. Start by importing wandb:

import wandb

Then, initialize wandb with wandb.init(). You can pass several optional parameters to wandb.init(), such as:

  • project: The name of the project where you’re logging runs. This could be any string, and a new project will be created if it doesn’t already exist.
  • entity: The username or team name under which the project is to be created.

An example of initializing Wandb for a project named β€˜my_project’ under username β€˜my_username’ would be:

wandb.init(project='my_project', entity='my_username')

After running wandb.init(), a new run will be created on the WanDB website, where you can track your model’s progress, visualize results, and more.

Let’s create a simple project that utilizes WandB (Weights and Biases) for experiment tracking. This project will be about classifying the CIFAR-10 dataset using a Convolutional Neural Network (CNN) implemented in Pytorch.

Project: CIFAR-10 Image Classification with Pytorch and WandB

Overview

The CIFAR-10 dataset consists of 60,000 32x32 color images in 10 classes, with 6,000 images per class. The goal of this project is to classify the images into these classes.

We will use a Convolutional Neural Network (CNN) in Pytorch to perform this classification task. The model’s performance will be logged and visualized using WandB, a tool for machine learning experiment tracking.

Steps

1. Setting Up the Environment

Install the necessary libraries.

!pip install torch torchvision wandb
Requirement already satisfied: torch in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (2.0.1)
Requirement already satisfied: torchvision in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (0.15.2)
Requirement already satisfied: wandb in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (0.16.4)
Requirement already satisfied: filelock in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from torch) (3.12.3)
Requirement already satisfied: typing-extensions in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from torch) (4.7.1)
Requirement already satisfied: sympy in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from torch) (1.12)
Requirement already satisfied: networkx in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from torch) (3.1)
Requirement already satisfied: jinja2 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from torch) (3.1.2)
Requirement already satisfied: numpy in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from torchvision) (1.25.2)
Requirement already satisfied: requests in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from torchvision) (2.31.0)
Requirement already satisfied: pillow!=8.3.*,>=5.3.0 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from torchvision) (10.0.0)
Requirement already satisfied: Click!=8.0.0,>=7.1 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (8.1.7)
Requirement already satisfied: GitPython!=3.1.29,>=1.0.0 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (3.1.36)
Requirement already satisfied: psutil>=5.0.0 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (5.9.5)
Requirement already satisfied: sentry-sdk>=1.0.0 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (1.40.6)
Requirement already satisfied: docker-pycreds>=0.4.0 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (0.4.0)
Requirement already satisfied: PyYAML in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (6.0.1)
Requirement already satisfied: setproctitle in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (1.3.3)
Requirement already satisfied: setuptools in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (68.2.2)
Requirement already satisfied: appdirs>=1.4.3 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (1.4.4)
Requirement already satisfied: protobuf!=4.21.0,<5,>=3.19.0 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from wandb) (4.24.3)
Requirement already satisfied: six>=1.4.0 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from docker-pycreds>=0.4.0->wandb) (1.16.0)
Requirement already satisfied: gitdb<5,>=4.0.1 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from GitPython!=3.1.29,>=1.0.0->wandb) (4.0.10)
Requirement already satisfied: charset-normalizer<4,>=2 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from requests->torchvision) (3.2.0)
Requirement already satisfied: idna<4,>=2.5 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from requests->torchvision) (3.4)
Requirement already satisfied: urllib3<3,>=1.21.1 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from requests->torchvision) (2.0.4)
Requirement already satisfied: certifi>=2017.4.17 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from requests->torchvision) (2024.2.2)
Requirement already satisfied: MarkupSafe>=2.0 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from jinja2->torch) (2.1.3)
Requirement already satisfied: mpmath>=0.19 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from sympy->torch) (1.3.0)
Requirement already satisfied: smmap<6,>=3.0.1 in /Users/ruangguru/Projects/ai-bootcamp/env/lib/python3.11/site-packages (from gitdb<5,>=4.0.1->GitPython!=3.1.29,>=1.0.0->wandb) (5.0.0)

[notice] A new release of pip is available: 23.3.2 -> 24.0
[notice] To update, run: pip install --upgrade pip

2. Import the Libraries

import torch
from torch import nn, optim
import torchvision
from torchvision import datasets, transforms
import wandb
import torch.nn.functional as F

3. Initialize WandB

wandb.login()
run = wandb.init(project='cifar10-classification', entity='ricky-kurniawan')
wandb: W&B API key is configured. Use `wandb login --relogin` to force relogin
Tracking run with wandb version 0.16.4
Run data is saved locally in /Users/ruangguru/Projects/ai-bootcamp/09_mlops/01_wandb/wandb/run-20240307_162118-scdpzfhm
Syncing run pretty-dew-44 to Weights & Biases (docs)
View project at https://wandb.ai/ricky-kurniawan/cifar10-classification
View run at https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/scdpzfhm

4. Prepare the Data

Load the CIFAR-10 dataset. Normalize the data and create dataloaders.

transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])

trainset = datasets.CIFAR10(root='./data', train=True, download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=64, shuffle=True)

testset = datasets.CIFAR10(root='./data', train=False, download=True, transform=transform)
testloader = torch.utils.data.DataLoader(testset, batch_size=64, shuffle=False)
Files already downloaded and verified
Files already downloaded and verified

5. Define the Model

Define a simple CNN model.

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(3, 6, 5)
        self.pool = nn.MaxPool2d(2, 2)
        self.conv2 = nn.Conv2d(6, 16, 5)
        self.fc1 = nn.Linear(16 * 5 * 5, 120)
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))
        x = x.view(-1, 16 * 5 * 5)
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x

model = Net()

6. Set Up the Loss Function and Optimizer

criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9)

7. Train the Model

Train the model and log the loss and accuracy to WandB.

def train_model(run, model, criterion, optimizer, trainloader):
    for epoch in range(10):  
        running_loss = 0.0
        correct_predictions = 0
        total_predictions = 0
        
        for i, data in enumerate(trainloader, 0):
            inputs, labels = data

            optimizer.zero_grad()

            outputs = model(inputs)
            loss = criterion(outputs, labels)
            loss.backward()
            optimizer.step()

            running_loss += loss.item()

            _, predicted = torch.max(outputs.data, 1)
            total_predictions += labels.size(0)
            correct_predictions += (predicted == labels).sum().item()

            if i % 200 == 199:    # Every 200 mini-batches
                print('[Epoch %d, Mini-batch %5d] Loss: %.3f' % (epoch + 1, i + 1, running_loss / 2000))
                wandb.log({'Epoch': epoch + 1, 'Loss': running_loss / 2000})
                wandb.log({'Epoch': epoch + 1, 'Loss': running_loss / 2000, 'Accuracy': correct_predictions / total_predictions * 100})
                running_loss = 0.0
                correct_predictions = 0
                total_predictions = 0

train_model(run, model, criterion, optimizer, trainloader)
[Epoch 1, Mini-batch   200] Loss: 0.230
[Epoch 1, Mini-batch   400] Loss: 0.230
[Epoch 1, Mini-batch   600] Loss: 0.229
[Epoch 2, Mini-batch   200] Loss: 0.218
[Epoch 2, Mini-batch   400] Loss: 0.209
[Epoch 2, Mini-batch   600] Loss: 0.200
[Epoch 3, Mini-batch   200] Loss: 0.193
[Epoch 3, Mini-batch   400] Loss: 0.189
[Epoch 3, Mini-batch   600] Loss: 0.183
[Epoch 4, Mini-batch   200] Loss: 0.179
[Epoch 4, Mini-batch   400] Loss: 0.175
[Epoch 4, Mini-batch   600] Loss: 0.171
[Epoch 5, Mini-batch   200] Loss: 0.166
[Epoch 5, Mini-batch   400] Loss: 0.163
[Epoch 5, Mini-batch   600] Loss: 0.161
[Epoch 6, Mini-batch   200] Loss: 0.155
[Epoch 6, Mini-batch   400] Loss: 0.153
[Epoch 6, Mini-batch   600] Loss: 0.151
[Epoch 7, Mini-batch   200] Loss: 0.148
[Epoch 7, Mini-batch   400] Loss: 0.145
[Epoch 7, Mini-batch   600] Loss: 0.145
[Epoch 8, Mini-batch   200] Loss: 0.141
[Epoch 8, Mini-batch   400] Loss: 0.139
[Epoch 8, Mini-batch   600] Loss: 0.142
[Epoch 9, Mini-batch   200] Loss: 0.137
[Epoch 9, Mini-batch   400] Loss: 0.136
[Epoch 9, Mini-batch   600] Loss: 0.135
[Epoch 10, Mini-batch   200] Loss: 0.133
[Epoch 10, Mini-batch   400] Loss: 0.132
[Epoch 10, Mini-batch   600] Loss: 0.129

8. Evaluate the Model

Evaluate the model on the test data and log the test accuracy to WandB.

correct = 0
total = 0
with torch.no_grad():
    for data in testloader:
        images, labels = data
        outputs = model(images)
        _, predicted = torch.max(outputs.data, 1)
        total += labels.size(0)
        correct += (predicted == labels).sum().item()

print('Accuracy of the network on the test images: %d %%' % (100 * correct / total))

wandb.log({'Test Accuracy': 100.0*correct/total})
Accuracy of the network on the test images: 52 %

This section of the code calculates the accuracy of the model on the test set and logs this test accuracy to WandB for visualization and tracking.

At this point, you can go to the WandB website, navigate to your project, and see a live visualization of your model’s loss and accuracy throughout the training process. This helps to understand how well the model is learning and provides insights for further improvements.

Finally, don’t forget to close your WandB run after you’re done:

run.finish()

Run history:


Accuracy β–β–‚β–‚β–ƒβ–ƒβ–„β–„β–…β–…β–…β–…β–†β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ
Epoch β–β–β–β–β–‚β–‚β–‚β–‚β–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–„β–„β–„β–„β–…β–…β–…β–…β–†β–†β–†β–†β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆ
Loss β–ˆβ–ˆβ–ˆβ–ˆβ–‡β–‡β–‡β–†β–…β–…β–…β–…β–„β–„β–„β–„β–„β–„β–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–β–β–β–β–β–
Test Accuracy ▁

Run summary:


Accuracy 53.6875
Epoch 10
Loss 0.12919
Test Accuracy 52.62

View run pretty-dew-44 at: https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/scdpzfhm
View job at https://wandb.ai/ricky-kurniawan/cifar10-classification/jobs/QXJ0aWZhY3RDb2xsZWN0aW9uOjE0NjA0NTAwNg==/version_details/v3
Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
Find logs at: ./wandb/run-20240307_162118-scdpzfhm/logs

This ensures all resources are properly freed and all logs are uploaded to the WandB server. This step is crucial to make sure all your model training progress and metrics are properly saved and can be reviewed later in the WandB dashboard.

Basic Concepts of WanDB

Here’s what the WanDB dashboard looks like:

A. Projects

In Wandb, a project is a collection of related machine learning experiments (known as β€œruns”). It provides a shared space where you and your team can compare results, share insights, and discuss potential improvements. Each project has a dedicated page on the Wandb web application, showcasing visualizations, comparisons, and other useful metrics. We set the project name when we do wandb.init() in Step-3 above.

B. Runs

A run in Wandb is a single execution of your machine learning script. During a run, you can log various metrics, such as loss and accuracy, system performance data, and even media like images or 3D objects. Each run gets its page in the Wandb web application, where you can view and analyze logged data.

As you can see from the screenshot of the WanDB dashboard, on the left we have 5 runs in our project.

C. Artifacts

Artifacts in Wandb are used to handle version control of datasets, models, and other result files from runs. They help to track the inputs and outputs of your runs, providing a clear and useful lineage of your models and data. For example, an input artifact could be your training dataset, while output artifacts could be your trained model or predictions.

On the left you can click on the Artifacts navigation icon which will take you to:

In this case, the artifact is a Jupyter notebook file (job-git_github.com_ruang-guru_ai-bootcamp.git_09_mlops_01_wandb_00_wandb_intro_setup.ipynb) from the project cifar10-classification owned by the user ricky-kurniawan. The version of the artifact is specified after the colon - in this case, v1 indicates it’s the first version.

The artifact.download() command is used to download the artifact to the local machine for use in the current run.

In the β€œUsed By” section, the listed items are the runs that have used this artifact. For example, the run proud-salad-5 used this artifact. Information such as the run’s performance metrics, the project, the user, the artifact used, and the time of artifact creation is displayed.

In this case, run-yqpg9z5p-history:v0 is an output artifact of the run. This run history artifact contains information about the run, such as the logged metrics. This artifact is created automatically by W&B when you log metrics or other information during a run. This allows you to revisit the specifics of a run, analyze the performance, and potentially identify areas for improvement or further exploration.

D. Sweep

Sweep is a feature in Wandb for hyperparameter optimization. A sweep involves a set of runs, each with different hyperparameters, allowing you to explore a range of possibilities and identify the best parameters for your model. Wandb automates this process, generating a set of permutations of hyperparameters (based on a configuration file you create), running them, and logging the results. This makes it easier to optimize your model’s performance.

Let’s try doing a Sweep using our CIFAR-10 Project.

1. Setting Up the Configuration

First, we need to create a configuration for our sweep. This configuration will specify the range and distribution of hyperparameters for the sweep. Here’s a basic example:

sweep_config = {
    'method': 'random', #grid, random
    'metric': {
      'name': 'accuracy',
      'goal': 'maximize'   
    },
    'parameters': {
        'epochs': {
            'values': [5, 10, 15]
        },
        'batch_size': {
            'values': [16, 32, 64]
        },
        'learning_rate': {
            'min': 1e-5,
            'max': 0.1
        },
    }
}

In this configuration, we’re specifying that we want to use a random search method (other options are grid for grid search and bayes for Bayesian optimization), and that our goal is to maximize accuracy. We’re also specifying the range of values for the hyperparameters that we want to optimize: epochs, batch size, and learning rate.

2. Initialize the Sweep

sweep_id = wandb.sweep(sweep_config, project="cifar10-classification")
Create sweep with ID: kqyufn28
Sweep URL: https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28

This command initializes the sweep and returns a sweep ID. This ID uniquely identifies the sweep in WandB.

3. Define the Train Function

Next, we need to modify the training function to accept configurations and log them to WandB. Add the following lines at the beginning of the function:

# Define a global counter
global_counter = 0
max_runs = 5

def train():
    global global_counter
    if global_counter >= max_runs:
        return
    global_counter += 1
    with wandb.init(config=sweep_config):
        config = wandb.config
        model = Net() 
        criterion = nn.CrossEntropyLoss()
        optimizer = optim.SGD(model.parameters(), lr=config.learning_rate, momentum=0.9)
        train_model(run, model, criterion, optimizer, trainloader)

In this function, wandb.init() is called with the sweep configuration to start a new run. wandb.config is then used to access the hyperparameters for the current run.

For the purpose of teaching, we are limiting the run to a maksimum of 5. Naturally you should let the sweep run and try out all possible combinations which may take a long time. You can safely remove all lines containing β€œglobal_counter” for real case study.

4. Run the Sweep

wandb.agent(sweep_id, train)
wandb: Agent Starting Run: jbxkaalg with config:
wandb:  batch_size: 64
wandb:  epochs: 10
wandb:  learning_rate: 0.052303023093622406
wandb: Currently logged in as: ricky-kurniawan. Use `wandb login --relogin` to force relogin
Tracking run with wandb version 0.16.4
Run data is saved locally in /Users/ruangguru/Projects/ai-bootcamp/09_mlops/01_wandb/wandb/run-20240307_162427-jbxkaalg
Syncing run gentle-sweep-1 to Weights & Biases (docs)
Sweep page: https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View project at https://wandb.ai/ricky-kurniawan/cifar10-classification
View sweep at https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View run at https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/jbxkaalg 
[Epoch 1, Mini-batch   200] Loss: 0.205
[Epoch 1, Mini-batch   400] Loss: 0.171
[Epoch 1, Mini-batch   600] Loss: 0.161
[Epoch 2, Mini-batch   200] Loss: 0.152
[Epoch 2, Mini-batch   400] Loss: 0.148
[Epoch 2, Mini-batch   600] Loss: 0.145
[Epoch 3, Mini-batch   200] Loss: 0.141
[Epoch 3, Mini-batch   400] Loss: 0.141
[Epoch 3, Mini-batch   600] Loss: 0.140
[Epoch 4, Mini-batch   200] Loss: 0.137
[Epoch 4, Mini-batch   400] Loss: 0.136
[Epoch 4, Mini-batch   600] Loss: 0.137
[Epoch 5, Mini-batch   200] Loss: 0.131
[Epoch 5, Mini-batch   400] Loss: 0.134
[Epoch 5, Mini-batch   600] Loss: 0.132
[Epoch 6, Mini-batch   200] Loss: 0.130
[Epoch 6, Mini-batch   400] Loss: 0.132
[Epoch 6, Mini-batch   600] Loss: 0.136
[Epoch 7, Mini-batch   200] Loss: 0.127
[Epoch 7, Mini-batch   400] Loss: 0.134
[Epoch 7, Mini-batch   600] Loss: 0.135
[Epoch 8, Mini-batch   200] Loss: 0.129
[Epoch 8, Mini-batch   400] Loss: 0.132
[Epoch 8, Mini-batch   600] Loss: 0.129
[Epoch 9, Mini-batch   200] Loss: 0.125
[Epoch 9, Mini-batch   400] Loss: 0.130
[Epoch 9, Mini-batch   600] Loss: 0.132
[Epoch 10, Mini-batch   200] Loss: 0.127
[Epoch 10, Mini-batch   400] Loss: 0.128
[Epoch 10, Mini-batch   600] Loss: 0.135

Run history:


Accuracy β–β–„β–…β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–‡β–‡β–‡β–‡β–ˆβ–‡β–‡β–ˆβ–‡β–‡β–ˆβ–‡β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‡
Epoch β–β–β–β–β–‚β–‚β–‚β–‚β–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–„β–„β–„β–„β–…β–…β–…β–…β–†β–†β–†β–†β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆ
Loss β–ˆβ–ˆβ–…β–„β–ƒβ–ƒβ–ƒβ–ƒβ–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–β–β–‚β–‚β–β–β–‚β–‚β–β–β–‚β–‚β–β–β–‚β–β–β–β–β–‚β–β–β–β–‚

Run summary:


Accuracy 54.84375
Epoch 10
Loss 0.13498

View run gentle-sweep-1 at: https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/jbxkaalg
Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
Find logs at: ./wandb/run-20240307_162427-jbxkaalg/logs 
wandb: Agent Starting Run: hil12twe with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.0778152359691495
Tracking run with wandb version 0.16.4
Run data is saved locally in /Users/ruangguru/Projects/ai-bootcamp/09_mlops/01_wandb/wandb/run-20240307_162717-hil12twe
Syncing run balmy-sweep-2 to Weights & Biases (docs)
Sweep page: https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View project at https://wandb.ai/ricky-kurniawan/cifar10-classification
View sweep at https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View run at https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/hil12twe 
[Epoch 1, Mini-batch   200] Loss: 0.206
[Epoch 1, Mini-batch   400] Loss: 0.184
[Epoch 1, Mini-batch   600] Loss: 0.173
[Epoch 2, Mini-batch   200] Loss: 0.162
[Epoch 2, Mini-batch   400] Loss: 0.164
[Epoch 2, Mini-batch   600] Loss: 0.164
[Epoch 3, Mini-batch   200] Loss: 0.168
[Epoch 3, Mini-batch   400] Loss: 0.161
[Epoch 3, Mini-batch   600] Loss: 0.163
[Epoch 4, Mini-batch   200] Loss: 0.160
[Epoch 4, Mini-batch   400] Loss: 0.161
[Epoch 4, Mini-batch   600] Loss: 0.161
[Epoch 5, Mini-batch   200] Loss: 0.159
[Epoch 5, Mini-batch   400] Loss: 0.162
[Epoch 5, Mini-batch   600] Loss: 0.166
[Epoch 6, Mini-batch   200] Loss: 0.164
[Epoch 6, Mini-batch   400] Loss: 0.166
[Epoch 6, Mini-batch   600] Loss: 0.169
[Epoch 7, Mini-batch   200] Loss: 0.164
[Epoch 7, Mini-batch   400] Loss: 0.166
[Epoch 7, Mini-batch   600] Loss: 0.167
[Epoch 8, Mini-batch   200] Loss: 0.171
[Epoch 8, Mini-batch   400] Loss: 0.169
[Epoch 8, Mini-batch   600] Loss: 0.169
[Epoch 9, Mini-batch   200] Loss: 0.166
[Epoch 9, Mini-batch   400] Loss: 0.173
[Epoch 9, Mini-batch   600] Loss: 0.174
[Epoch 10, Mini-batch   200] Loss: 0.169
[Epoch 10, Mini-batch   400] Loss: 0.172
[Epoch 10, Mini-batch   600] Loss: 0.173

Run history:


Accuracy β–β–„β–†β–‡β–‡β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‡β–‡β–ˆβ–ˆβ–‡β–‡β–‡β–‡β–ˆβ–‡β–‡β–‡β–‡β–‡
Epoch β–β–β–β–β–‚β–‚β–‚β–‚β–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–„β–„β–„β–„β–…β–…β–…β–…β–†β–†β–†β–†β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆ
Loss β–ˆβ–ˆβ–…β–ƒβ–β–β–‚β–‚β–‚β–‚β–β–‚β–β–β–β–β–β–β–β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–ƒβ–ƒβ–‚β–‚β–‚β–‚β–ƒβ–ƒβ–‚β–‚β–ƒβ–ƒ

Run summary:


Accuracy 40.67188
Epoch 10
Loss 0.17274

View run balmy-sweep-2 at: https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/hil12twe
Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
Find logs at: ./wandb/run-20240307_162717-hil12twe/logs 
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: wtl4q1ha with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.010806903248724136
Tracking run with wandb version 0.16.4
Run data is saved locally in /Users/ruangguru/Projects/ai-bootcamp/09_mlops/01_wandb/wandb/run-20240307_163023-wtl4q1ha
Syncing run worldly-sweep-3 to Weights & Biases (docs)
Sweep page: https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View project at https://wandb.ai/ricky-kurniawan/cifar10-classification
View sweep at https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View run at https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/wtl4q1ha 
[Epoch 1, Mini-batch   200] Loss: 0.224
[Epoch 1, Mini-batch   400] Loss: 0.190
[Epoch 1, Mini-batch   600] Loss: 0.167
[Epoch 2, Mini-batch   200] Loss: 0.146
[Epoch 2, Mini-batch   400] Loss: 0.141
[Epoch 2, Mini-batch   600] Loss: 0.136
[Epoch 3, Mini-batch   200] Loss: 0.126
[Epoch 3, Mini-batch   400] Loss: 0.124
[Epoch 3, Mini-batch   600] Loss: 0.120
[Epoch 4, Mini-batch   200] Loss: 0.114
[Epoch 4, Mini-batch   400] Loss: 0.111
[Epoch 4, Mini-batch   600] Loss: 0.113
[Epoch 5, Mini-batch   200] Loss: 0.105
[Epoch 5, Mini-batch   400] Loss: 0.104
[Epoch 5, Mini-batch   600] Loss: 0.104
[Epoch 6, Mini-batch   200] Loss: 0.099
[Epoch 6, Mini-batch   400] Loss: 0.099
[Epoch 6, Mini-batch   600] Loss: 0.099
[Epoch 7, Mini-batch   200] Loss: 0.091
[Epoch 7, Mini-batch   400] Loss: 0.092
[Epoch 7, Mini-batch   600] Loss: 0.095
[Epoch 8, Mini-batch   200] Loss: 0.087
[Epoch 8, Mini-batch   400] Loss: 0.088
[Epoch 8, Mini-batch   600] Loss: 0.091
[Epoch 9, Mini-batch   200] Loss: 0.081
[Epoch 9, Mini-batch   400] Loss: 0.085
[Epoch 9, Mini-batch   600] Loss: 0.086
[Epoch 10, Mini-batch   200] Loss: 0.079
[Epoch 10, Mini-batch   400] Loss: 0.082
[Epoch 10, Mini-batch   600] Loss: 0.083

Run history:


Accuracy β–β–ƒβ–„β–…β–…β–…β–†β–†β–†β–†β–‡β–†β–‡β–‡β–‡β–‡β–‡β–‡β–‡β–ˆβ–‡β–ˆβ–ˆβ–‡β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ
Epoch β–β–β–β–β–‚β–‚β–‚β–‚β–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–„β–„β–„β–„β–…β–…β–…β–…β–†β–†β–†β–†β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆ
Loss β–ˆβ–ˆβ–†β–…β–„β–„β–„β–„β–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–β–β–β–‚β–β–β–β–β–β–β–β–

Run summary:


Accuracy 70.96094
Epoch 10
Loss 0.083

View run worldly-sweep-3 at: https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/wtl4q1ha
Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
Find logs at: ./wandb/run-20240307_163023-wtl4q1ha/logs 
wandb: Agent Starting Run: ettfrm5h with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.0795927437531589
Tracking run with wandb version 0.16.4
Run data is saved locally in /Users/ruangguru/Projects/ai-bootcamp/09_mlops/01_wandb/wandb/run-20240307_163323-ettfrm5h
Syncing run sparkling-sweep-4 to Weights & Biases (docs)
Sweep page: https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View project at https://wandb.ai/ricky-kurniawan/cifar10-classification
View sweep at https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View run at https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/ettfrm5h 
[Epoch 1, Mini-batch   200] Loss: 0.201
[Epoch 1, Mini-batch   400] Loss: 0.178
[Epoch 1, Mini-batch   600] Loss: 0.174
[Epoch 2, Mini-batch   200] Loss: 0.166
[Epoch 2, Mini-batch   400] Loss: 0.168
[Epoch 2, Mini-batch   600] Loss: 0.166
[Epoch 3, Mini-batch   200] Loss: 0.166
[Epoch 3, Mini-batch   400] Loss: 0.162
[Epoch 3, Mini-batch   600] Loss: 0.166
[Epoch 4, Mini-batch   200] Loss: 0.161
[Epoch 4, Mini-batch   400] Loss: 0.162
[Epoch 4, Mini-batch   600] Loss: 0.165
[Epoch 5, Mini-batch   200] Loss: 0.163
[Epoch 5, Mini-batch   400] Loss: 0.161
[Epoch 5, Mini-batch   600] Loss: 0.163
[Epoch 6, Mini-batch   200] Loss: 0.164
[Epoch 6, Mini-batch   400] Loss: 0.163
[Epoch 6, Mini-batch   600] Loss: 0.162
[Epoch 7, Mini-batch   200] Loss: 0.163
[Epoch 7, Mini-batch   400] Loss: 0.168
[Epoch 7, Mini-batch   600] Loss: 0.162
[Epoch 8, Mini-batch   200] Loss: 0.164
[Epoch 8, Mini-batch   400] Loss: 0.164
[Epoch 8, Mini-batch   600] Loss: 0.165
[Epoch 9, Mini-batch   200] Loss: 0.168
[Epoch 9, Mini-batch   400] Loss: 0.168
[Epoch 9, Mini-batch   600] Loss: 0.172
[Epoch 10, Mini-batch   200] Loss: 0.172
[Epoch 10, Mini-batch   400] Loss: 0.170
[Epoch 10, Mini-batch   600] Loss: 0.168

Run history:


Accuracy β–β–…β–†β–†β–†β–‡β–‡β–‡β–‡β–ˆβ–‡β–‡β–ˆβ–ˆβ–ˆβ–‡β–ˆβ–ˆβ–ˆβ–‡β–ˆβ–ˆβ–ˆβ–ˆβ–‡β–‡β–‡β–‡β–‡β–ˆ
Epoch β–β–β–β–β–‚β–‚β–‚β–‚β–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–„β–„β–„β–„β–…β–…β–…β–…β–†β–†β–†β–†β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆ
Loss β–ˆβ–ˆβ–„β–ƒβ–‚β–‚β–‚β–‚β–‚β–‚β–β–‚β–β–β–β–‚β–β–β–β–β–‚β–‚β–β–β–β–β–‚β–β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–ƒβ–ƒβ–ƒβ–ƒβ–‚

Run summary:


Accuracy 42.27344
Epoch 10
Loss 0.168

View run sparkling-sweep-4 at: https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/ettfrm5h
Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
Find logs at: ./wandb/run-20240307_163323-ettfrm5h/logs 
wandb: Agent Starting Run: sli7lpw0 with config:
wandb:  batch_size: 64
wandb:  epochs: 10
wandb:  learning_rate: 0.02990947205127064
Tracking run with wandb version 0.16.4
Run data is saved locally in /Users/ruangguru/Projects/ai-bootcamp/09_mlops/01_wandb/wandb/run-20240307_163623-sli7lpw0
Syncing run earnest-sweep-5 to Weights & Biases (docs)
Sweep page: https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View project at https://wandb.ai/ricky-kurniawan/cifar10-classification
View sweep at https://wandb.ai/ricky-kurniawan/cifar10-classification/sweeps/kqyufn28
View run at https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/sli7lpw0 
[Epoch 1, Mini-batch   200] Loss: 0.207
[Epoch 1, Mini-batch   400] Loss: 0.169
[Epoch 1, Mini-batch   600] Loss: 0.157
[Epoch 2, Mini-batch   200] Loss: 0.139
[Epoch 2, Mini-batch   400] Loss: 0.137
[Epoch 2, Mini-batch   600] Loss: 0.134
[Epoch 3, Mini-batch   200] Loss: 0.125
[Epoch 3, Mini-batch   400] Loss: 0.123
[Epoch 3, Mini-batch   600] Loss: 0.125
[Epoch 4, Mini-batch   200] Loss: 0.116
[Epoch 4, Mini-batch   400] Loss: 0.118
[Epoch 4, Mini-batch   600] Loss: 0.117
[Epoch 5, Mini-batch   200] Loss: 0.110
[Epoch 5, Mini-batch   400] Loss: 0.113
[Epoch 5, Mini-batch   600] Loss: 0.113
[Epoch 6, Mini-batch   200] Loss: 0.105
[Epoch 6, Mini-batch   400] Loss: 0.109
[Epoch 6, Mini-batch   600] Loss: 0.110
[Epoch 7, Mini-batch   200] Loss: 0.101
[Epoch 7, Mini-batch   400] Loss: 0.104
[Epoch 7, Mini-batch   600] Loss: 0.109
[Epoch 8, Mini-batch   200] Loss: 0.099
[Epoch 8, Mini-batch   400] Loss: 0.105
[Epoch 8, Mini-batch   600] Loss: 0.101
[Epoch 9, Mini-batch   200] Loss: 0.096
[Epoch 9, Mini-batch   400] Loss: 0.102
[Epoch 9, Mini-batch   600] Loss: 0.103
[Epoch 10, Mini-batch   200] Loss: 0.093
[Epoch 10, Mini-batch   400] Loss: 0.101
[Epoch 10, Mini-batch   600] Loss: 0.104

Run history:


Accuracy β–β–„β–„β–…β–…β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–‡β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–‡β–ˆβ–‡β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‡
Epoch β–β–β–β–β–‚β–‚β–‚β–‚β–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–ƒβ–„β–„β–„β–„β–…β–…β–…β–…β–†β–†β–†β–†β–†β–†β–†β–†β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆ
Loss β–ˆβ–ˆβ–†β–…β–„β–„β–„β–„β–ƒβ–ƒβ–ƒβ–ƒβ–‚β–‚β–ƒβ–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–‚β–β–β–‚β–‚β–β–β–‚β–‚β–β–β–‚β–‚

Run summary:


Accuracy 64.03125
Epoch 10
Loss 0.10403

View run earnest-sweep-5 at: https://wandb.ai/ricky-kurniawan/cifar10-classification/runs/sli7lpw0
Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
Find logs at: ./wandb/run-20240307_163623-sli7lpw0/logs 
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: rtnk6ct2 with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.04013296976488706
wandb: Agent Starting Run: bqo62g9l with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.019177756444346203
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 4xg5m9q5 with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.08786538663827181
wandb: Agent Starting Run: hccvb2aj with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.09598692806072671
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 9l2kwop6 with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.02399552057250517
wandb: Agent Starting Run: kxjxgcoz with config:
wandb:  batch_size: 64
wandb:  epochs: 10
wandb:  learning_rate: 0.001797124637780443
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 0hg7ms36 with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.039904043596502
wandb: Agent Starting Run: edi3pf6o with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.05230665673262373
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: b0rhoirl with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.07634240948572275
wandb: Agent Starting Run: b3kwtw9q with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.01182721332560279
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 8384njvv with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.03237160620647384
wandb: Agent Starting Run: 1s1lmbyu with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.006803383495698264
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: hmillp2c with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.08159207853619327
wandb: Agent Starting Run: 3qhajxuy with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.034446125391791926
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 47338itd with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.027886490017400307
wandb: Agent Starting Run: n1vn81r5 with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.06457802029413433
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: cgale92f with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.09884957347400788
wandb: Agent Starting Run: b4dzya1c with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.023560550660514372
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: xc1w68kc with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.0677273337735739
wandb: Agent Starting Run: 3qc17zvp with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.020872029295613256
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: bycum4ax with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.0009502174683948892
wandb: Agent Starting Run: ea13uxm4 with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.03875932439917796
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: cv9ymo3e with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.03643050258260523
wandb: Agent Starting Run: rdudww8g with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.07521550538459543
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: j0t2l6xn with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.0926547142606578
wandb: Agent Starting Run: hjs4wpup with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.03103738494870234
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: a1zb8bt0 with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.08247355964474885
wandb: Agent Starting Run: 6w9frll1 with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.04725600655064744
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: zncziwff with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.043450210349017096
wandb: Agent Starting Run: we8x72vm with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.06386487328353212
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: waxypahb with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.042724514638329464
wandb: Agent Starting Run: joi5zjsj with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.019430254130361313
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: xx88hfkr with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.07547491698020012
wandb: Agent Starting Run: kjhokwe7 with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.00019885257735111447
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: snjtvnuw with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.01991671460892379
wandb: Agent Starting Run: a8335nge with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.01795560439131162
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: d8qd3tqr with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.04822296635022568
wandb: Agent Starting Run: oynds7y0 with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.044998018963751635
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 9ll122fe with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.07340241778955124
wandb: Agent Starting Run: c0vlax52 with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.002283546818615913
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: bt6cj0un with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.022668240334927293
wandb: Agent Starting Run: mrdz8w9o with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.05708704882360202
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: umbfbt8n with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.08304445886730834
wandb: Agent Starting Run: 3toy7vas with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.08614471640332713
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: c6esvwxi with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.00975734470018658
wandb: Agent Starting Run: h3jc09p7 with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.0006792195004773297
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: zm9y7f8f with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.035608281690541546
wandb: Agent Starting Run: flc4dlyi with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.0747376446577179
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: d6zjnt74 with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.0147811451253074
wandb: Agent Starting Run: curacom8 with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.03078741176536328
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: dz7jp1uz with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.03647067772864392
wandb: Agent Starting Run: 88b26sl8 with config:
wandb:  batch_size: 64
wandb:  epochs: 10
wandb:  learning_rate: 0.03164228010599915
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 1vx8j98a with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.060367573514628875
wandb: Agent Starting Run: 0uh1isax with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.09465268340140094
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: houw5pqz with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.04899465514091969
wandb: Agent Starting Run: w2o09wx9 with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.07761409080397712
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 6kmdrt5z with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.03191007075831756
wandb: Agent Starting Run: eyedp7sg with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.0389982675078401
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: knj1pboz with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.05755347056457952
wandb: Agent Starting Run: uxyumcsf with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.08216783864067907
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: f31dej07 with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.08075061578012568
wandb: Agent Starting Run: fs20edf9 with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.07661993280501622
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 78j6qti3 with config:
wandb:  batch_size: 64
wandb:  epochs: 10
wandb:  learning_rate: 0.07472375363565593
wandb: Agent Starting Run: 0rbbjcnf with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.062440351458098856
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 0p5dph5h with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.095907049984872
wandb: Agent Starting Run: ftrbpjom with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.03250778308923327
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: xtyertyc with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.08278359995673532
wandb: Agent Starting Run: gujy3vad with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.07482464086791861
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 7xsa7jdq with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.07774704322127651
wandb: Agent Starting Run: zoodx6p9 with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.008214304338534106
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: wlhlk1y9 with config:
wandb:  batch_size: 64
wandb:  epochs: 10
wandb:  learning_rate: 0.06960814841631531
wandb: Agent Starting Run: 5hv0lxmk with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.060997982236647166
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 5ejtdrgs with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.029460761958973136
wandb: Agent Starting Run: udyjpx2s with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.03138232430215001
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: qmhjuyl3 with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.026927982306497504
wandb: Agent Starting Run: kqe4dcbu with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.0905052326377824
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 48petedk with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.08917413284685244
wandb: Agent Starting Run: 9cve8j4b with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.09537012996979832
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: cbm95fdg with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.07043152434511156
wandb: Agent Starting Run: o7m7zvkl with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.04866494456570549
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: gjpy9r0r with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.07551029408789373
wandb: Agent Starting Run: 91fp8jt4 with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.09478472201588872
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 2pu84sxr with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.017711354426403886
wandb: Agent Starting Run: e1sgeidh with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.09046440046876088
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: k25dk7xy with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.091976710481526
wandb: Agent Starting Run: c1klbar5 with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.03743326691366536
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: qhqtolza with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.08201836927894077
wandb: Agent Starting Run: 10tr16u4 with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.08041062549078154
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: inlf4ymj with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.087283456718445
wandb: Agent Starting Run: 819rpaa4 with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.038763119130549355
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: odxiv32v with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.01556679894974132
wandb: Agent Starting Run: bh9pdy8l with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.04817863423238899
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: y38z5nr6 with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.016870862546940594
wandb: Agent Starting Run: y9kzkujv with config:
wandb:  batch_size: 64
wandb:  epochs: 15
wandb:  learning_rate: 0.06284544026485823
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: kmwm24tu with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.07492284565203791
wandb: Agent Starting Run: i6dl9cis with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.0427534134153622
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: txbvsz69 with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.05326299600938453
wandb: Agent Starting Run: ccw4uhmi with config:
wandb:  batch_size: 32
wandb:  epochs: 10
wandb:  learning_rate: 0.0480183490301355
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: ivkyuvs8 with config:
wandb:  batch_size: 64
wandb:  epochs: 10
wandb:  learning_rate: 0.06865922913712455
wandb: Agent Starting Run: ci2r2guu with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.07737784852558242
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 2pnrxxz0 with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.05050030078259987
wandb: Agent Starting Run: rq54hcia with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.0910912726664158
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 00clda8n with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.07824430646201114
wandb: Agent Starting Run: uri7slp8 with config:
wandb:  batch_size: 16
wandb:  epochs: 10
wandb:  learning_rate: 0.06346494197531323
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 3u5pu7dy with config:
wandb:  batch_size: 32
wandb:  epochs: 15
wandb:  learning_rate: 0.09495262177733783
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: ajuxu8d8 with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.02070714388688412
wandb: Agent Starting Run: xlz9f0ad with config:
wandb:  batch_size: 16
wandb:  epochs: 15
wandb:  learning_rate: 0.029069817807907232
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: yhcoekh6 with config:
wandb:  batch_size: 16
wandb:  epochs: 5
wandb:  learning_rate: 0.08499498160510198
wandb: Agent Starting Run: simg5ivw with config:
wandb:  batch_size: 64
wandb:  epochs: 5
wandb:  learning_rate: 0.034975887586655756
wandb: Sweep Agent: Waiting for job.
wandb: Job received.
wandb: Agent Starting Run: 90gbjyps with config:
wandb:  batch_size: 32
wandb:  epochs: 5
wandb:  learning_rate: 0.08929397988150486
wandb: Sweep Agent: Waiting for job.
wandb: Sweep Agent: Exiting.

This command will start the sweep we just defined. WandB will call the train function with the different combinations of hyperparameters defined in the sweep configuration.

Here’s what the Sweep Dashboard looks like:

We can clearly see the which combination is more effective by looking at the charts, we can then dive in to fine tune our hyperparameters.

WandB’s sweeps are a powerful tool for optimizing your model’s hyperparameters. By integrating WandB with your model training code, you can automate the process of training many models with different hyperparameters, and then easily compare their performance on the WandB’s dashboard.

You can view the full project at this link: CIFAR10 WanDB Dashboard

Back to top