Frequently Asked Questions
With NimbleBox, it is easy to use train models uninterrupted. However, there are a few steps to take before training models on NimbleBox.
Ensure that you have an active internet connection.
The Dedicated Hardware is enabled.
You have chosen a sufficient amount of RAM.
When you run in a terminal, be sure to use the tmux command. Note: We recommend you save models intermittently. This allows you to restore the most recently saved version and resume training.
No. We understand and respect your need for confidentiality in your models, data, and code. We care about our customers' privacy and data security. We do not share any of your data with others. So, we give you complete control over whom you can invite and share your instance.
Welcome to NimbleBox.You can use the lowest configuration CPU to upload data, code, and debug. This method can help reduce costs.We recommend preemptive projects if you train small models and do not mind restarting the computing instance and continuing work.If you are training large models for many hours, check the FAQ 'How can I train the model on NimbleBox uninterruptedly?'You can find further information on hardware configuration in the documentation. See this twitter thread to learn more on choosing the hardware
We currently support popular IDEs such as Jupyter, Jupyter Lab, and Visual Studio Code. If you have other IDE requests, let us know over the chat inside the app.
Yes, you can increase the auto-shutdown timer when running the instance. However, changing the auto-shutdown time will not stop or restart the instance.To increase the auto-shutdown timer, open your instance in the workspace. Then, click Edit in the Auto-Shutdown section and change the auto-shutdown time.
Your instance comes with around 8GB of pre-installed libraries. However, you can increase your storage and pay only for your use.
There are multiple ways to upload a file on NimbleBox. It depends on where your file is present. If your data is currently in your local, we recommend using 'File Manager' to upload the file. You can use Google integrations and GitHub to upload your files to NimbleBox. You can use the Kaggle API method to download your files directly from Kaggle to NimbleBox. Use the curl command to get your data from the URL directly on NimbleBox.To unzip a file on NimbleBox: The file type .zip is recommended for uploading large datasets on NimbleBox. You can install the unzip library and use it.On NimbleBox, you unzip the file by going to Custom ports, selecting the Default Ports drop-down menu, and choosing VS Code.In the VS Code terminal, type the following commands:sudo apt install unzip unzip
Yes. NimbleBox Backup has a powerful and easy-to-use file backup system that lets you have peace of mind while protecting your important files. In addition, backup can be helpful in the event of data loss.
It is easy to restore the data. From Build, Go to the 'Settings' of your instance and select the 'Restore Now' option to restore the latest data.
Yes, your data is safe. Resetting your environment does not affect your data or code. It will only remove all the user-installed libraries.A reset environment is helpful if your current environment is messed-up due to version incompatibility and you want to return to the default packages given by NimbleBox.
Yes, the instance has live monitoring, which displays the amount of RAM you use at all times.Additionally, you can type the following command in the terminal to check your instance's total, free and available RAM. watch -n 0.2 free -m
Pre-emptive projects are intended for use in the short term. Pre-emptible projects cost significantly less than dedicated instances.We recommend pre-emptive projects for short-term projects and quick edits to code and noncritical work that does not require more than 24 hours of active time.
If your instance has stalled at Staging for more than 5 minutes, you will see a 'Force Stop' option. You can use this option to stop the instance and restart it again.
No. Deletion of a instance will permanently delete the instance from the workspace.However, if you have already created a backup, you can restore the most recent backup and continue working on your instance.
With NimbleBox free tier, you can gain hands-on experience with NimbleBox services and pay only for what you use.You can use the free trial for the first organizational workspace you create.Things that you can try out in the free trial.Create a instance and train a model.Deploy the app.Create a team and invite people.
In order to start the instance, you need to provide your billing address and payment details. The 14-day-free-trial gets activated once you have provided your card details.
NimbleBox uses the card holder's name, billing address, and zip code to verify that an authorized cardholder is the person placing the order. This process prevents fraudulent credit card transactions and ensures everyone's safety. A nominal amount of $0.50 is added to your bank account, then credited back to your account.If you have any doubts, please email us at email@example.com or chat with us on live chat.
No. Your subscription will automatically renew monthly and be charged to your account. However, you will not be charged for the usage if you haven't used any NimbleBox resources (Pay-as-you-go).
On the NimbleBox dashboard, choose 'Billing' on the side navigation menu.To cancel your subscription, click the 'Subscriptions' tab. Then, in the plan details, click 'Cancel Subscription.'We confirm that the existing subscription will be canceled at the end of your current subscription cycle by clicking cancel subscriptions.
NimbleBox free tier, you can gain hands-on experience with NimbleBox services. The subscription fee for a 14-days free trial will not be charged. However, you are required to pay for the resources you have utilized at the end of the free trial.
Kernel and Packages
We currently support TensorFlow 2.7.0. When using any specific version of TensorFlow, make sure it is compatible with the CUDA version on NimbleBox.
NimbleBox comes with the following packages installed in your project by default.
Yes. Any package installed using pip install
The installation of libraries on NimbleBox is simple. You can install the libraries by running pip install
When RAM fills up, the kernel often restarts. To monitor the amount of RAM dynamically, you can use the watch -n 0.2 free -m command. You can also upgrade your RAM configuration or modify your code to resolve this issue.
NimbleBox stores your files on the cloud and doesn't depend on your local computer path. NimbleBox is similar to a Linux environment, and all Linux commands are suitable inside the terminal. The pwd command is the go-to command for displaying the current folder's path.
No worries! Just reset the environment by clicking on 'Reset Environment' from the build settings. The reset environment helps to return to the default packages provided by NimbleBox. For further information on resetting the environment, read the docs.
nbox is SDK for NimbleBox.ai. It provides built-in access to all the APIs and packages them in the most user-friendly manner. You need to install nbox to deploy models and schedule them timely (jobs).You can install nbox using the following command.pip install nbox.For more information about the nbox package and usage, see this documentation.
That's right. NimbleBox Deploy is cloud-agnostic so that you can deploy models regardless of the cloud provider (such as AWS, GCP, Azure, and more)
Nimblebox makes deploying and monitoring your ML models in production a breeze.NimbleBox supports the following frameworks:TensorFlow,sklearn,Haiku.If you're interested in using a framework not currently supported, contact us at firstname.lastname@example.org.
The autoscaling feature allows you to scale your Instance Application as its demand increases or decreases. As a result, you can save money by decreasing the number of resources used when demand is low and adding resources when demand is high to prevent performance issues.
You can create rules that automatically add or remove pods based on CPU usage by setting a minimum and a maximum number of pods.For any customizable scaling options, reach out to us via chat.
When a new deployment fails, the system reverts to the previous model. The previous model is removed only when the new model has been deployed successfully. The system evicts the running deployments and uses the latest successful model deployed.
To see the model logs, from the 'Deployment' page in the workspace, go to the 'Models' tab and select the 'View Models' tab.
If you are a Maintainer or an Editor of Deployment, you can delete the deployment group. By doing so, all models in the Deployment will be deleted. Unfortunately, there is no way to retrieve back the files that have been deleted. To delete a deployment group, follow these steps:Go to your deployment page in the NimbleBox workspace. Select 'Manage' on the top-right corner of the deployment group you want to delete.On the Settings tab, scroll down to the Danger Zone section and click the Delete button.Type the name of the deployment group you wish to delete, and click 'Delete Group' to confirm the deletion.
API key provides security to ensure that no third party is accessing the URL.Helps you to share model access privately.Embed your API key in your Instance Application (mobile app or website) and make the model available through it.
Your current running model is called a pinned model. Any new uploaded model will be a pinned model.
Here is how you can roll back to the model.Select the model you want to roll back and click ‘run.’
No worries! You can roll back to the previous model. View all the models in the deployment and select 'run' from the model you want to roll back.
Once the model is deployed, and the access key is generated, copy the cURL command below the access key and run it in the terminal. This will provide you with all the information (metadata) about the model.
Jobs help automate redundant tasks and schedule the job at a particular interval.For more information on jobs and their usage, check this documentation.
Run is a single execution of a job from start to end.
To view your jobs' run logs, go to the Jobs page in your workspace, select the jobs you want to check, and click 'View Run Logs.'
The Job Flow feature lets you see how jobs are executed in NimbleBox. Go to your NimbleBox workspace, and from the Sidebar bar, select Jobs. Click on the job flow tab to view your job's execution flow.
You can view the history of your jobs from the jobs workspace page. Just click on the 'Job Logs' tab to see the old job tasks of the respective job.
When you're logged into your workspace and on the jobs page, click on the trigger button beside each job.
Sometimes in practice, your job might be failing because of a bug in the code. 'Pausing' helps to avoid any more failing job runs and resume once the bug is fixed.
The Workspace is a centralized place to collaborate, build, deploy and automate. You can invite team members and collaborate on Workspace. Learn more about the workspaces here.
Bring your data and files to NimbleBox from where it is stored. Connect to third-party apps such as google drive / Github and download the files directly to NimbleBox.Github Integration.Create an instance with the Github repository or clone a repository to NimbleBox.Google Drive.Connect to google drive and easily browse and upload your files to Nimblebox.
NimbleBox free trial is applicable only once to gain hands-on experience on the product. So, you get a 14-day free trial of the NimbleBox account for the first Workspace you create, but not for any subsequent ones.
No, there can be only one Owner in the Workspace. Check out our documentation for more information about 'roles and permissions.'
To remove a member from a group, go to the 'Groups' tab. You will be able to view the list of members from the drop-down.From the member's menu, choose 'Remove Members' and confirm. To learn more about roles and permissions, click here.
If the Owner hasn't transferred ownership and left the company, you can reset the password from the Owner's email address. After that, you can log in with the Owner's account and transfer the ownership. If the Owner's email is deactivated, you can reach out to our support team via chat and provide details on the Workspace and the new Owner, and a member of our support team will get back to you right away.
You can contact your workspace Owner or Admin and request them to add your account with the appropriate roles and permissions.For more information about roles and permissions, click here.
We would love to hear feedback from you. Head over to the chat button in the app and give feedback. Our skilled product team checks feedback regularly and takes steps to improve the product 😊
Unfortunately, support doesn’t have a phone number due to security reasons. However, you can reach out to our friendly staff in chat, and we will be happy to help :)
Our friendly support team is just one chat away. Reach out to the team in the chat icon available at the bottom right of the app.
We support all the issues related to NimbleBox product, account, and billing :)