zoom_in Click to enlarge

Kaggle Kernels

▲ ▼

description Kaggle Kernels Overview

Kaggle Kernels provides a simplified, browser-based Jupyter Notebook environment directly within the Kaggle platform. Its ideal for beginners and those new to data science, offering pre-configured environments and access to a vast library of datasets and notebooks. While limited in customization, its a fantastic way to get started.

help Kaggle Kernels FAQ

Does Kaggle Kernels come with free GPU access?

Yes, Kaggle provides free access to GPUs including NVIDIA T4 and P100 instances for running your notebooks directly in the browser. You can enable GPU acceleration in the notebook settings, though weekly GPU hour quotas apply and individual sessions are time-limited.

What are the runtime limits for Kaggle Kernels?

Kaggle Kernels CPU sessions are limited to 12 hours per run, while GPU sessions are also capped at 12 hours per session with a weekly GPU quota typically around 30 hours. These constraints encourage efficient code but are generous enough for most data science experimentation and Kaggle competitions.

Can I use my own datasets in a Kaggle Kernel?

Yes, you can upload private datasets to Kaggle and access them directly within your Kernel alongside the thousands of public datasets already hosted on the platform. Private datasets remain visible only to you and collaborators you explicitly share them with.

Can I install custom Python packages in a Kaggle Notebook?

You can install additional packages using pip or conda directly within a notebook cell, just like in a local Jupyter environment. However, since environments reset between sessions, you will need to include your installation commands at the top of your notebook each time.

swap_horiz

Looking for Kaggle Kernels alternatives? Compare top competitors ranked & scored

arrow_forward