wandb · mintlify · May 1, 2026 · May 1, 2026
@@ -1,4 +1,5 @@
 ---
 title: "W&B Tutorials & Blog"
+description: "Browse W&B tutorials, articles, and guides on the Fully Connected blog covering ML experiment tracking, MLOps, and LLM workflows."
 url: "https://wandb.ai/fully-connected"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "W&B Courses"
+description: "Take free W&B platform courses to learn experiment tracking, hyperparameter sweeps, model management, and LLM evaluations."
 url: "https://wandb.ai/site/courses/?e-filter-32fa548-course-category=wb-platform"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "API Reference"
+description: "Browse the W&B Inference API reference for the OpenAI-compatible chat completions endpoint and supported parameters."
 url: "/inference/api-reference"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "Support: Inference"
+description: "Browse W&B Inference support articles covering API errors, authentication, billing, rate limits, and troubleshooting."
 url: "/support/inference"
 ---
@@ -1,6 +1,5 @@
 ---
-description:
-  Create, track, and use a dataset artifact with W&B.
+description: "Follow this tutorial to create, track, and use a dataset artifact in W&B and learn artifact versioning fundamentals."
 title: "Tutorial: Create, track, and use a dataset artifact"
 ---
 

@@ -1,4 +1,5 @@
 ---
 title: "CLI"
+description: "Browse the W&B Command Line Interface (CLI) reference for logging in, running jobs, executing sweeps, and more."
 url: "/models/ref/cli"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "Public API"
+description: "Browse the W&B Public API Python reference for querying runs, projects, artifacts, reports, and other resources programmatically."
 url: "/models/ref/python/public-api"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "Python SDK"
+description: "Browse the W&B Python SDK API reference including installation, classes, functions, and data types."
 url: "/models/ref/python"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "Query Expression Language"
+description: "Browse the W&B Query Expression Language reference for selecting and aggregating data in custom panels."
 url: "/models/ref/query-panel"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "Reports and Workspaces API"
+description: "Browse the Reports and Workspaces API reference for creating and managing reports and workspaces programmatically."
 url: "/models/ref/wandb_workspaces/reports"
 ---
@@ -1,6 +1,6 @@
 ---
 title: Delete runs
-description: Delete runs from a W&B project using the W&B App or the Public API, and learn how deleted run data is removed from storage on Dedicated Cloud and Self-Managed deployments.
+description: "Delete W&B runs through the W&B App or Public API, and learn how deleted run data is removed from storage on Dedicated Cloud and Self-Managed deployments."
 ---
 
 ## Delete runs

@@ -1,4 +1,5 @@
 ---
 title: "Support: Models"
+description: "Browse W&B Models support articles covering experiments, sweeps, artifacts, reports, billing, and account management."
 url: "/support/models"
 ---
@@ -1,6 +1,6 @@
 ---
 title: Manage bucket storage and costs
-description: Understand how W&B uses object storage, how deletion maps to bucket bytes, and how to reduce usage on self-managed, Dedicated Cloud, and bring-your-own-bucket deployments.
+description: "How W&B uses object storage, how deletion maps to bucket bytes, and how to reduce usage on self-managed, Dedicated Cloud, and BYOB deployments."
 ---
 
 When you use [Bring your own bucket (BYOB)](/platform/hosting/data-security/secure-storage-connector), [W&B Dedicated Cloud](/platform/hosting/hosting-options/dedicated-cloud), or [W&B Self-Managed](/platform/hosting/hosting-options/self-managed), your team often pays cloud storage providers directly. This page explains what occupies your bucket, how W&B removes objects after deletion in the app or API, and what you should expect in practice.

@@ -1,6 +1,6 @@
 ---
 title: W&B Mobile App (iOS)
-description: From your iPhone or iPad, track training runs, review console logs, view line plots, search project panels with regular expressions, and explore your W&B Models projects.
+description: "Track training runs, review console logs, view line plots, search projects with regex, and explore W&B Models projects from your iPhone or iPad."
 ---
 
 <Columns cols={4}>

@@ -1,5 +1,6 @@
 ---
 title: Are there best practices for using Launch effectively?
+description: "Best practices for using W&B Launch effectively, including queue setup, service accounts, and managing hyperparameters."
 ---
 
 1. Create the queue before starting the agent to enable easy configuration. Failure to do this results in errors that prevent the agent from functioning until a queue is added.

@@ -1,5 +1,6 @@
 ---
 title: I do not want W&B to build a container for me, can I still use Launch?
+description: "Use a pre-built Docker image with W&B Launch instead of letting Launch build a container for you."
 ---
 
 To launch a pre-built Docker image, execute the following command. Replace the placeholders in the `<>` with your specific information:

@@ -1,5 +1,6 @@
 ---
 title: I do not like clicking- can I use Launch without going through the UI?
+description: "Use the W&B Launch CLI subcommand to launch and manage jobs without using the W&B App UI."
 ---
 
 Yes. The standard `wandb` CLI includes a `launch` subcommand to launch jobs. For more information, run:

@@ -1,5 +1,6 @@
 ---
 title: How do I control who can push to a queue?
+description: "Restrict who can push to a W&B Launch queue by managing membership of the queue's owning team."
 ---
 
 Queues are specific to a user team. Define the owning entity during queue creation. To restrict access, modify team membership.
@@ -1,6 +1,6 @@
 ---
-title: When multiple jobs in a Docker queue download the same artifact, is any caching
-  used, or is it re-downloaded every run?
+title: "When multiple jobs in a Docker queue download the same artifact, is any caching used, or is it re-downloaded every run?"
+description: "Configure shared artifact caching for Docker Launch queues to avoid redundant downloads across parallel jobs."
 ---
 
 No caching exists. Each launch job operates independently. Configure the queue or agent to mount a shared cache using Docker arguments in the queue configuration.

@@ -1,5 +1,6 @@
 ---
 title: Can I specify a Dockerfile and let W&B build a Docker image for me?
+description: "Specify a Dockerfile so W&B Launch builds the Docker image for you, useful for stable requirements with changing code."
 ---
 This feature suits projects with stable requirements but frequently changing codebases.
 

@@ -1,6 +1,6 @@
 ---
-title: Can Launch automatically provision (and spin down) compute resources for me
-  in the target environment?
+title: "Can Launch automatically provision (and spin down) compute resources for me in the target environment?"
+description: "Whether W&B Launch can automatically provision and tear down compute resources in SageMaker, Vertex, and Kubernetes."
 ---
 
 This process depends on the environment. Resources provision in Amazon SageMaker and Vertex. In Kubernetes, autoscalers automatically adjust resources based on demand. Solution Architects at W&B assist in configuring Kubernetes infrastructure to enable retries, autoscaling, and the use of spot instance node pools. For support, contact support@wandb.com or use your shared Slack channel.
@@ -1,5 +1,6 @@
 ---
 title: How does W&B Launch build images?
+description: "How W&B Launch builds container images based on the job source, accelerator base image, and queue configuration."
 ---
 
 The steps for building an image depend on the job source and the specified accelerator base image in the resource configuration.

@@ -1,6 +1,6 @@
 ---
-title: Is `wandb launch -d` or `wandb job create image` uploading a whole docker artifact
-  and not pulling from a registry?
+title: "Is `wandb launch -d` or `wandb job create image` uploading a whole docker artifact and not pulling from a registry?"
+description: "Whether `wandb launch -d` and `wandb job create image` upload Docker artifacts, and how to push images to a registry."
 ---
 
 No, the `wandb launch -d` command does not upload images to a registry. Upload images to a registry separately. Follow these steps:

@@ -1,6 +1,6 @@
 ---
-title: Does Launch support parallelization?  How can I limit the resources consumed
-  by a job?
+title: "Does Launch support parallelization? How can I limit the resources consumed by a job?"
+description: "How W&B Launch supports parallelization across GPUs and nodes, and how to limit resources consumed per job."
 ---
 
 Launch supports scaling jobs across multiple GPUs and nodes. Refer to the [Volcano integration guide](/models/integrations) for details.

@@ -1,5 +1,6 @@
 ---
 title: How do I make W&B Launch work with Tensorflow on GPU?
+description: "Configure a custom accelerator base image so W&B Launch jobs run TensorFlow with GPU support."
 ---
 
 For TensorFlow jobs using GPUs, specify a custom base image for the container build. This ensures proper GPU utilization during runs. Add an image tag under the `builder.accelerator.base_image` key in the resource configuration. For example:

@@ -1,5 +1,6 @@
 ---
 title: How do I fix a "permission denied" error in Launch?
+description: "Troubleshoot the \"Launch Error: Permission denied\" error caused by missing W&B login or insufficient project permissions."
 ---
 
 If you encounter the error message `Launch Error: Permission denied`, it indicates insufficient permissions to log to the desired project. Possible causes include:

@@ -1,5 +1,6 @@
 ---
 title: What permissions does the agent require in Kubernetes?
+description: "Required Kubernetes role and permissions for running the W&B Launch agent, including pods, configmaps, secrets, and logs."
 ---
 
 The following Kubernetes manifest creates a role named `wandb-launch-agent` in the `wandb` namespace. This role allows the agent to create pods, configmaps, secrets, and access pod logs in the `wandb` namespace. The `wandb-cluster-role` enables the agent to create pods, access pod logs, create secrets, jobs, and check job status across any specified namespace.
@@ -1,5 +1,6 @@
 ---
 title: What requirements does the accelerator base image have?
+description: "Requirements for the W&B Launch accelerator base image, including OS compatibility, GPU drivers, and CUDA support."
 ---
 
 For jobs utilizing an accelerator, provide a base image that includes the necessary accelerator components. Ensure the following requirements for the accelerator image:

@@ -1,5 +1,6 @@
 ---
 title: How can admins restrict which users have modify access?
+description: "How W&B team admins use queue config templates to restrict which queue fields non-admin users can modify."
 ---
 
 Control access to certain queue fields for users who are not team administrators through [queue config templates](/platform/launch/setup-queue-advanced/). Team administrators define which fields non-admin users can view, and set the editing limits. Only team administrators have the ability to create or edit queues.
@@ -1,6 +1,6 @@
 ---
-title: Can you specify secrets for jobs/automations? For instance, an API key which
-  you do not wish to be directly visible to users?
+title: "Can you specify secrets for jobs/automations? For instance, an API key which you do not wish to be directly visible to users?"
+description: "Use Kubernetes secrets to pass API keys to W&B Launch jobs and automations without exposing them in run configs."
 ---
 
 Yes. Follow these steps:

@@ -1,5 +1,6 @@
 ---
 title: launch api
+description: "Reference for the W&B Launch `launch` Python API used to programmatically run a Launch experiment from a job source."
 ---
 
 <Card title="View the source code" href="https://www.github.com/wandb/wandb/tree/v0.20.1/wandb/sdk/launch/_launch.py#L249-L331" icon="github"/>

@@ -1,5 +1,6 @@
 ---
 title: launch_add
+description: "Reference for the W&B Launch `launch_add` Python API used to enqueue a Launch experiment from a URI, job, or Docker image."
 ---
 
 <Card title="View the source code" href="https://www.github.com/wandb/wandb/tree/v0.20.1/wandb/sdk/launch/_launch_add.py#L34-L131" icon="github"/>

@@ -1,5 +1,6 @@
 ---
 title: LaunchAgent
+description: "Reference for the W&B Launch `LaunchAgent` Python class used to poll run queues and launch runs."
 ---
 
 <Card title="View the source code" href="https://www.github.com/wandb/wandb/tree/v0.20.1/wandb/sdk/launch/agent/agent.py#L164-L924" icon="github"/>

@@ -1,4 +1,5 @@
 ---
 title: "W&B Pricing"
+description: "Compare W&B pricing plans for individuals, teams, and enterprises across W&B Models, Weave, Inference, Training, and Sandboxes."
 url: "https://wandb.ai/site/pricing/"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "W&B Inference"
+description: "Open-source foundation model inference as a service from W&B with OpenAI-compatible APIs and built-in tracing through Weave."
 url: "/inference"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "W&B Models"
+description: "Track ML experiments, hyperparameter sweeps, datasets, and model versions with the W&B Models MLOps platform."
 url: "/models"
 ---
@@ -1,5 +1,6 @@
 ---
 title: "W&B Sandboxes"
+description: "Spin up isolated, on-demand compute sandboxes for running ML code, agents, and experiments in W&B-managed containers."
 url: "/sandboxes"
 tag: Preview
 ---
@@ -1,4 +1,5 @@
 ---
 title: "W&B Training"
+description: "Run serverless reinforcement learning and supervised fine-tuning jobs on managed GPU infrastructure with W&B Training."
 url: "/training"
 ---
@@ -1,4 +1,5 @@
 ---
 title: "W&B Weave"
+description: "Trace, evaluate, and monitor LLM applications and AI agents with the W&B Weave observability and evaluation platform."
 url: "/weave"
 ---
@@ -1,5 +1,6 @@
 ---
 title: W&B Sandboxes
+description: "W&B Sandboxes provide on-demand, isolated compute environments you can create, run code in, and discard programmatically."
 mode: wide
 ---
 

@@ -1,6 +1,6 @@
 ---
 title: Create sandboxes
-description: Learn how to create W&B Sandboxes.
+description: "Create W&B Sandboxes to run ML code and agents in isolated, on-demand compute containers with their own filesystem, network, and process space."
 ---
 
 <Warning>

@@ -1,6 +1,6 @@
 ---
 title: Secrets
-description: Learn how to manage secrets in W&B Sandboxes.
+description: "Manage secrets in W&B Sandboxes by using the W&B Secrets Manager to securely access API keys, tokens, and credentials inside a sandbox container."
 ---
 
 <Warning>

@@ -1,4 +1,5 @@
 ---
 title: "W&B Platform Security"
+description: "Learn how W&B protects customer data with SOC 2 compliance, encryption, access controls, and platform security best practices."
 url: "https://wandb.ai/site/security/"
 ---
@@ -1,5 +1,6 @@
 ---
 title: Support Home
+description: "Search support articles, product documentation, and community resources for W&B Models, Weave, Inference, and the platform."
 mode: "center"
 ---
 
@@ -41,22 +42,22 @@ and the W&B community.
 ### W&B Models
 
 <Card title="How can I configure the name of the run in my training code?" href="/support/models/articles/how-can-i-configure-the-name-of-the-run-" arrow="true" horizontal>
-  At the beginning of the training script, call wandb.init with an experiment name. For example: wandb.init(name="my_aweso ...
+  Learn how to configure the name of the run in your training code in W&B.
 
 <Badge stroke shape="pill" color="orange" size="md">[Experiments](/support/models/tags/experiments)</Badge> 
 </Card>
 <Card title="How do I use custom CLI commands with sweeps?" href="/support/models/articles/how-do-i-use-custom-cli-commands-with-sw" arrow="true" horizontal>
-  You can use W&B Sweeps with custom CLI commands if training configuration passes command-line arguments. In the example  ...
+  Learn how to use custom CLI commands with sweeps in W&B: You can use W&B Sweeps with custom CLI commands if training configuration passes command-line...
 
 <Badge stroke shape="pill" color="orange" size="md">[Sweeps](/support/models/tags/sweeps)</Badge> 
 </Card>
 <Card title="Is it possible to save metrics offline and sync them to W&B later?" href="/support/models/articles/is-it-possible-to-save-metrics-offline-a" arrow="true" horizontal>
-  By default, wandb.init starts a process that syncs metrics in real time to the cloud. For offline use, set two environme ...
+  By default, `wandb.init` starts a process that syncs metrics in real time to the cloud.
 
 <Badge stroke shape="pill" color="orange" size="md">[Experiments](/support/models/tags/experiments)</Badge> <Badge stroke shape="pill" color="orange" size="md">[Environment Variables](/support/models/tags/environment-variables)</Badge> <Badge stroke shape="pill" color="orange" size="md">[Metrics](/support/models/tags/metrics)</Badge> 
 </Card>
 <Card title="What does wandb.init do to my training process?" href="/support/models/articles/what-does-wandbinit-do-to-my-training-pr" arrow="true" horizontal>
-  When wandb.init() runs in a training script, an API call creates a run object on the servers. A new process starts to st ...
+  When `wandb.init()` runs in a training script, an API call creates a run object on the servers.
 
 <Badge stroke shape="pill" color="orange" size="md">[Environment Variables](/support/models/tags/environment-variables)</Badge> <Badge stroke shape="pill" color="orange" size="md">[Experiments](/support/models/tags/experiments)</Badge> 
 </Card>

@@ -1,5 +1,6 @@
 ---
 title: "API error code 401 - Authentication failed"
+description: "Resolve the W&B Inference 401 \"Authentication failed\" error caused by incorrect API credentials, project entity, or project name."
 keywords: ["Authentication & Access"]
 ---
 

@@ -1,5 +1,6 @@
 ---
 title: "API error code 402 - You exceeded your current quota"
+description: "Resolve the W&B Inference 402 quota error caused by running out of credits or hitting your monthly spending cap."
 keywords: ["Quotas & Rate Limits", "Billing"]
 ---
 

@@ -1,5 +1,6 @@
 ---
 title: "API error code 403 - Country, region, or territory not supported"
+description: "Resolve the W&B Inference 403 error returned when accessing the API from a country, region, or territory that is not supported."
 keywords: ["Authentication & Access"]
 ---
 

@@ -1,5 +1,6 @@
 ---
 title: "API error code 403 - The inference gateway is not enabled for your organization"
+description: "Resolve the W&B Inference 403 error returned when the inference gateway is not enabled for your organization."
 keywords: ["Authentication & Access", "Administrator"]
 ---
 

@@ -1,5 +1,6 @@
 ---
 title: "API error code 429 - Concurrency limit reached for requests"
+description: "Resolve the W&B Inference 429 error returned when you exceed the concurrency limit for parallel API requests."
 keywords: ["Quotas & Rate Limits"]
 ---
 

@@ -1,5 +1,6 @@
 ---
 title: "API error code 500 - The server had an error while processing your request"
+description: "Resolve the W&B Inference 500 internal server error, a transient issue that usually clears with a retry."
 keywords: ["Server Errors"]
 ---
 

@@ -1,5 +1,6 @@
 ---
 title: "API error code 503 - The engine is currently overloaded"
+description: "Resolve the W&B Inference 503 \"engine overloaded\" error returned during periods of high traffic on the API."
 keywords: ["Server Errors"]
 ---