Migrating from Neptune to Weights & Biases

2026-02-20 · Source: Weights & Biases · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, extended

Summary

Weights & Biases (W&B) is offering a migration guide and support for users transitioning from Neptune AI, which is shutting down on March 5th. The migration process involves exporting data from Neptune to local storage, then importing it into W&B. Users need W&B and Neptune API keys, and must clone the Neptune migration script repository. The guide details setting up environment variables for API keys, using UV for dependency management, and executing command-line scripts for data export and import. It also covers visualizing rich media, with an option to clone a separate repository for in-platform visualization. A key difference highlighted is W&B's project-centric run structure and its `wb.log` method for metrics, configurations, and media, contrasting with Neptune's `append` and `upload` functions. W&B also offers automatic dashboards and sharable saved views, differing from Neptune's manual comparison views.

Key takeaway

For MLOps Engineers and Data Scientists managing experiment tracking, you should prioritize exporting your Neptune AI data to local storage immediately, well before the March 5th shutdown. Leverage the provided migration scripts and W&B's support to transition your projects. Familiarize yourself with W&B's project-run structure and its unified `wb.log` method for metrics and media to ensure a smooth integration and avoid data loss.

Key insights

Migrate from Neptune AI to Weights & Biases by March 5th using provided scripts and a structured workflow.

Principles

• Prioritize data export before platform shutdown.
• Use environment variables for API key security.
• Understand platform-specific data logging paradigms.

Method

Clone the Neptune migration repo, install UV, set API keys as environment variables, export Neptune data using `neptune-exporter`, then import into W&B using `wandb-loader`, specifying entity and file paths.

In practice

• Export Neptune data locally immediately.
• Review Neptune projects to select data for migration.
• Use `wb.log` for all W&B metric and media logging.

Topics

• Neptune AI
• Weights & Biases
• Data Migration
• Experiment Tracking
• MLOps

Best for: Machine Learning Engineer, MLOps Engineer, Data Scientist

Related on AIssential

• [P] We made GoodSeed, a pleasant ML experiment tracker — GoodSeed is an open-source ML experiment tracker with a clean UI and Neptune migration capabilities.
• Oops, I Deleted It Again — AI coding agents, even when tasked with migration, can autonomously delete user data, necessitating careful oversight.
• Software Deployment for Machine Learning: Strategies That Work in Production — ML deployment fundamentally differs from software deployment due to models' probabilistic nature and data dependency, requiring continuous adaptation and specialized strategies.
• Migrate MLflow tracking servers to Amazon SageMaker AI with serverless MLflow — Migrating MLflow to SageMaker's serverless App reduces operational overhead and improves resource scaling.
• Monitor your training runs with the Weights & Biases mobile app — The W&B mobile app provides real-time AI experiment monitoring and alerts on iOS devices.
• The HackerNoon Newsletter: Vibe Coding Ends at Localhost (6/21/2026) — The tech landscape in June 2026 is marked by AI deployment hurdles, specialized database optimizations, and strategic acquisitions.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Weights & Biases.