Weights & Biases Documentation llms.txt · AI Docs Browser · mdream [llms.txt Tools](https://mdream.dev/llms-txt/wandb/llms-txt) # Weights & Biases Documentation llms.txt docs.wandb.ai/llms.txt [Raw llms.txt](https://docs.wandb.ai/llms.txt) [Raw llms-full.txt](https://docs.wandb.ai/llms-full.txt) Copy Markdown [](https://docs.wandb.ai/llms.txt) llms.txt Docs Use ARIA for autoresearch Chat with ARIA Governance and security Overview W&B Tutorials & Blog W&B Courses Example Code & Notebooks Get Started with Weights & Biases W&B HiveMind Build AI agents, applications, and models Serverless Inference API overview Chat completions List models Usage examples Model lifecycle Use Serverless LoRA Inference Available models Prerequisites API Reference Enable JSON mode View reasoning information Enable streaming responses Enable structured output Call tools Support: Inference Creating a fine-tuned LoRA Cline with Serverless Inference UI guide Usage information and limits W&B Models Console logs Manage workspace, section, and panel settings Custom charts overview Tutorial: Use custom charts Panels Bar plots Save and diff code Line plots overview Line plot reference Point aggregation Smooth line plots Media panels Parallel coordinates Parameter importance Query panels overview Embed objects Compare run metrics Scatter plots Keyboard shortcuts LEET terminal UI Artifacts overview Tutorial: Create, track, and use a dataset artifact Create an artifact Create an artifact alias Create an artifact version Artifact data privacy and compliance Delete an artifact Download and use artifacts Explore artifact lineage graphs Manage artifact storage and memory allocation Track external files Manage artifact data retention Update an artifact Automations overview Manage automations with the API Automation events and scopes Overview Create a Slack automation Create a webhook automation Tutorial: Project run-failure alert automation Tutorial: Registry artifact alias automation Automation tutorial overview View an automation's history Overview Compare runs with Eval Tables Create an Eval Table View Eval Tables Integrations overview Hugging Face Accelerate Add W&B to a Python library Azure OpenAI fine-tuning Hugging Face Diffusers Hugging Face Hugging Face Transformers Hydra PyTorch Ignite Keras PyTorch Lightning OpenAI API OpenAI fine-tuning OpenAI Gym PyTorch PyTorch Geometric Hugging Face Simple Transformers TensorFlow PyTorch torchtune XGBoost YOLOv5 YOLOX LLM Evaluation Jobs Evaluate a hosted API model Evaluate a model checkpoint Evaluation benchmark catalog Get Started with W&B Models W&B Quickstart Reference overview CLI Python SDK Query Expression Language Reports and Workspaces API CLI Reference SDK 0.28.0 wandb agent wandb artifact wandb artifact cache wandb artifact cache cleanup wandb artifact get wandb artifact ls wandb artifact put wandb beta wandb beta core wandb beta core start wandb beta core stop wandb beta sync wandb controller wandb disabled wandb docker wandb docker-run wandb enabled wandb init wandb job wandb job create wandb job describe wandb job list wandb leet wandb leet config wandb leet run wandb leet symon wandb login wandb offline wandb online wandb pull wandb purge-cache wandb restore wandb server wandb server start wandb server stop wandb status wandb sweep wandb sync wandb verify Python SDK 0.28.0 Automations overview Automation DoNothing MetricChangeFilter MetricThresholdFilter MetricZScoreFilter NewAutomation OnAddArtifactAlias OnAddArtifactTag OnAddCollectionTag OnCreateArtifact OnLinkArtifact OnRemoveArtifactTag OnRemoveCollectionTag OnRunMetric OnRunState OnUnlinkArtifact RunStateFilter SendNotification SendWebhook Custom Charts overview bar() confusion_matrix() histogram line() line_series() plot_table() pr_curve() roc_curve() scatter() Data Types overview Audio box3d() EvalTable Histogram Html Image Molecule Object3D Plotly Table Video Experiments overview Artifact Run Settings System Metrics Reference Global Functions overview agent() controller() finish() init() login() restore() setup() sweep() teardown() Public API overview AgentRuns Api ArtifactCollection ArtifactCollections ArtifactFiles Artifacts ArtifactType ArtifactTypes Automations BetaReport DownloadHistoryResult File Files IncompleteRunHistoryError Member Project ProjectArtifactCollections Projects Registry Reports Run RunArtifacts Runs Sweep Sweeps Team User Query Expression Language overview artifact artifactType artifactVersion audio-file bokeh-file boolean entity file float html-file image-file int joined-table molecule-file number object3D-file partitioned-table project pytorch-model-file run string table user video-file W&B SDK Python coding cheat sheet Artifacts Logging Registry Runs Reports Workspaces Registry overview Reference an artifact version with aliases Configure registry access Create a collection Create a registry Delete registry Download an artifact from a registry Lineage graphs and audit history Link an artifact version to a collection Organize versions with tags Annotate collections Find registry items Reports overview Clone and export reports Collaborate on reports Create a report Compare runs across projects Edit a report Embed a report Example reports Overview Send an alert Semantic run plot legends Pin and compare runs View runs in a project Delete runs Filter runs Fork a run Organize runs Visualize CoreWeave infrastructure alerts Initialize runs Move a run to a different project or team Resume a run Rewind a run Customize run colors Find and customize a run's ID or name Run states Search runs Stop runs Add labels to runs with tags View a specific run in a project Support: Models Sweeps overview Add W&B (wandb) to your code Overview Tutorial: Create sweep job from project Initialize a sweep Manage algorithms locally Parallelize agents Manage sweeps Signal handling and sweep runs Start a sweep agent Sweep configuration options Sweeps troubleshooting Learn more about sweeps Visualize sweep results Tutorial: Define, initialize, and run a sweep Tables overview Log tables Export table data Example tables Tutorial: Log tables, visualize and query data Visualize and analyze tables Experiments overview Configure experiments Create an experiment Environment variables Track Jupyter notebooks Logging at scale and performance Overview Customize log axes Log distributed training experiments Log models Log summary metrics Log tables Log media and objects Create and track plots from experiments Track CSV files with experiments Projects Import and export data Reproduce experiments View experiments results Settings Manage billing settings Manage email settings Manage storage Team settings Manage user settings Deployment options overview Data encryption in Dedicated Cloud Configure IP allowlisting for Dedicated Cloud Access BYOB using pre-signed URLs Configure private connectivity to Dedicated Cloud Bring your own bucket (BYOB) Configure environment variables Dedicated Cloud Export data from Dedicated Cloud Rate limits Supported Dedicated Cloud regions Multi-tenant Cloud W&B Self-Managed deployment overview Access management Manage your organization Manage access control for projects Advanced IAM configuration Automate user and team management Use federated identities with SDK Configure SSO with LDAP Identity and access management (IAM) Manage users, groups, and roles with SCIM Use service accounts to automate workflows Configure SSO with OIDC Manage bucket storage and costs Track user activity with audit logs W&B Mobile App (iOS) View organization activity Configure Slack alerts Disable automatic updates for W&B Server Deploy on Air-Gapped Kubernetes Deploy W&B with Kubernetes Operator Rate limits Reference architecture Self-Managed infrastructure requirements Update W&B license and version W&B Launch Add job to queue Create and deploy jobs Create a launch job Launch integration guides Manage job inputs Launch FAQ Are there best practices for using Launch effectively? I don't want W&B to build a container for me, can I still use Launch? I don't like clicking, can I use Launch without going through the UI? How do I control who can push to a queue? When multiple jobs in a Docker queue download the same artifact, is any caching used, or is it re-downloaded every run? Can I specify a Dockerfile and let W&B build a Docker image for me? Can Launch automatically provision (and spin down) compute resources for me in the target environment? How W&B Launch builds images Is wandb launch -d or wandb job create image uploading a whole Docker artifact and not pulling from a registry? Does Launch support parallelization? How can I limit the resources consumed by a job? How to make W&B Launch work with TensorFlow on GPU How do I fix a "permission denied" error in Launch? What permissions does the agent require in Kubernetes? What requirements does the accelerator base image have? How can admins restrict which users have modify access? Can you specify secrets for jobs or automations? For instance, an API key which you do not wish to be directly visible to users? launch api launch_add LaunchAgent Monitor launch queue Launch terms and concepts View launch jobs Set up Launch Set up launch agent Tutorial: Set up W&B Launch with Docker Tutorial: Set up W&B Launch on Kubernetes Tutorial: Set up W&B Launch on SageMaker Configure launch queue Tutorial: Set up W&B Launch on Vertex AI Create sweeps with W&B Launch Tutorial: W&B Launch basics Use the W&B MCP Server Manage secrets Use W&B Skills W&B Pricing Serverless Inference W&B Models Serverless Sandboxes Serverless Training W&B Weave Reference Release Notes Release policies and processes W&B SDK releases Supported W&B Server releases Archived W&B Server releases Weave SDK releases Serverless Sandboxes Create sandboxes File operations Tutorial: Invoke an agent in a sandbox Sandbox lifecycle Tutorial: Train a PyTorch model Configuration Core Discovery Exceptions Types Utilities Run commands in a sandbox Secrets W&B Platform Security Serverless Training API overview Create Chat Completion Create Chat Completion Health Check System Check Create Model Delete Model Delete Model Checkpoints List Model Checkpoints Log Create Rl Training Job Create Sft Training Job Get Training Job Get Training Job Events Available models Prerequisites API Reference Serverless SFT How to use Serverless SFT How to use Serverless RL Usage information and limits Use your trained models Support Home Support: Serverless Inference Administrator Authentication & Access Billing Quotas & Rate Limits Server Errors Support: W&B Models Academic Administrator Alerts Anonymous API API Keys Artifacts Authentication AWS Billing Charts Connectivity Environment Variables Experiments Hyperparameter Inference Logs Metrics Notebooks Org Management Outage Privacy Projects Python Reports Resuming Run Crashes Runs SDK Security Storage Sweeps Tables Team Management Teams TensorBoard User Management Workspace Workspaces Wysiwyg Support: W&B Weave Client Info Code Capture Data Capture Evaluation Performance System Info Trace Data UI Rendering W&B Weave Choose an agent integration What is Weave? Overview Introduction to Evaluations Introduction to Traces Use Weave with W&B Models Audio with Weave Chain Of Density Codegen Custom Model Cost DSPy Prompt Optimization Feedback Prod HuggingFace Dataset Evaluations Import from CSV Leaderboard Quickstart Multi-agent structured output Not Diamond custom routing Trace and Evaluate a Computer Vision Pipeline with Weave Online Monitoring PII Data Handling Scorers as guardrails Use the Weave Service API to trace Quickstart: Set up custom agent observability Limits and expected behaviors Support: Weave Collect and track datasets Configure Weave environment variables Evaluations overview Compare and rank models Log media Track application versions with models Create prompt objects Store and track versions of prompts Set up automations Use builtin scorers Set up custom monitors Create dynamic Leaderboards in Evaluations Log evaluation data from your code Export evaluation data Set up guardrails Monitor using built-in signals Scoring overview Evaluate using local scorers Integrations overview Claude Agent SDK Claude Code plugin Codex plugin Google ADK OpenAI Agents SDK OpenClaw plugin Pi extension Agno Anthropic AutoGen Control automatic LLM call tracking Microsoft Azure Bedrock Bedrock Agents Cerebras Cohere CrewAI DSPy Google Groq Haystack Hugging Face Hub Instructor TypeScript SDK: third-party integration guide Koog LangChain LiteLLM LlamaIndex Local Models Model Context Protocol (MCP) and Weave MistralAI NVIDIA NIM OpenAI OpenAI Realtime API OpenRouter PydanticAI Smolagents Together AI Vercel AI SDK Verdict Verifiers Deployment options and security features Configure ingest sampling for Weave Self-Managed Manage Weave projects Set up a self-managed W&B Weave instance Define and log attributes Map columns in datasets Compare traces and other logged information Compare model performance using the Evaluation Playground Use the Playground to experiment with prompts Create and manage saved views Use Weave with W&B training runs Reference media in your own bucket (BYOB) using agent spans Set up annotation queues Review items in an annotation queue Reference media in your own bucket (BYOB) using Weave Op Call schema reference Track costs Trace your code Collect feedback and use annotations Get a handle to the Call object during execution Track and version objects Customize Ops Send OpenTelemetry Traces to Weave Query and export Calls Redact PII from traces Set Call display name Trace threads Trace agent integrations Trace your agents Set attributes and events on agent spans Batch logging for your agent Disable tracing Trace generator functions Use trace plots Trace sub-agents Link a W&B run to trace function calls Navigate the Weave Trace view Understand Ops, Calls, and Traces Update and delete Calls View agent activity Monitor your agents with signals View and customize trace display Write-ahead log Quickstart: Track LLM inputs and outputs Learn Weave with Serverless Inference Python SDK Service API TypeScript SDK Overview weave feedback op util weave_client trace_server_interface remote_http_trace_server Service API overview Export Genai Trace Genai Agent Versions Query Genai Agents Query Genai Conversation Chat Genai Conversation Spans Genai Custom Attrs Schema Genai Search Genai Spans Query Genai Spans Stats Genai Traces Chat Annotation Queue Add Calls Annotation Queue Create Annotation Queue Delete Annotation Queue Item Progress Update Annotation Queue Items Query Annotation Queue Read Annotation Queue Update Annotation Queues Query Stream Annotation Queues Stats Call End Call Read Call Start Call Start Batch Call Stats Call Update Calls Complete Calls Delete Calls Query Stats Calls Query Stream Calls Usage Trace Usage Cost Create Cost Purge Cost Query Dataset Create Dataset Delete Dataset List Dataset Read Eval Results Query Evaluation Run Create Evaluation Run Delete Evaluation Run Finish Evaluation Run List Evaluation Run Read Evaluate Model Evaluation Create Evaluation Delete Evaluation List Evaluation Read Evaluation Status Rescore Evaluation Feedback Aggregate Feedback Create Feedback Create Batch Feedback Payload Schema Feedback Purge Feedback Query Feedback Replace Feedback Stats File Content File Create Files Stats Image Create Inference Analysis Artificialanalysis Models Inference Catalog Models Inference Modelsdev Models Inference Router Openrouter Models Nvidia Hardware Model Create Model Delete Model List Model Read Aliases List Obj Add Tags Obj Create Obj Delete Obj Read Obj Remove Aliases Obj Remove Tags Obj Set Aliases Objs Query Tags List Export Trace Op Create Op Delete Op List Op Read Prediction Create Prediction Delete Prediction Finish Prediction List Prediction Read Refs Read Batch Link To Registry Scorer Create Scorer Delete Scorer List Scorer Read Calls Score Score Create Score Delete Score List Score Read Get Caller Location Projects Info Read Root Read Version Server Info Table Create Table Create From Digests Table Query Table Query Stats Table Query Stats Batch Table Update Threads Query Stream weave Conversation Dataset Evaluation EvaluationLogger LLM MessagesPrompt ObjectRef ScoreLogger StringPrompt SubAgent Tool Turn WeaveAdkPlugin WeaveClient WeaveObject createOpenAIAgentsTracingProcessor createOtelExtension endConversation endLLM endSession endTurn flushOTel getCurrentConversation getCurrentLLM getCurrentSession getCurrentTurn init instrumentOpenAIAgents login op patchRealtimeSession requireCurrentCallStackEntry requireCurrentChildSummary runIsolated startConversation startLLM startSession startSubagent startTool startTurn weaveAudio weaveImage withAttributes wrapClaudeAgentSdk wrapGoogleGenAI wrapOpenAI CallSchema CallsFilter ConversationInit GetAgentsOptions GetAgentSpansOptions GetAgentTurnOptions GetAgentTurnsOptions GetAgentVersionsOptions GetCallsOptions HttpResponse HTTPValidationError LLMInit Message Query Reasoning SortBy SubAgentInit ToolInit TurnInit Usage WeaveAudio WeaveImage Agent AgentMessage AgentSpan AgentTurn AgentVersion GetAgentSpansResult GetAgentsResult GetAgentTurnResult GetAgentTurnsResult GetAgentVersionsResult MessagePart Modality Op OpDecorator Response Role Session SessionInit Settings Build an evaluation Evaluate RAG applications Trace nested functions Tutorial: App versioning OpenAPI Specs openapi ## What is Weights & Biases Documentation's llms.txt? Weights & Biases Documentation publishes an `/llms.txt` file that provides AI systems with a structured index of its documentation. It follows the [llms.txt specification](https://llmstxt.org), organizing links to guides, API references, and tutorials under section headings so that LLMs like ChatGPT, Claude, and Gemini can quickly understand what Weights & Biases Documentation offers and where to find details. Pass Spec 2 Sections 893 Links 98.1 KB Size ~24.9k Tokens Sections: Docs · OpenAPI Specs ## Add Weights & Biases Documentation Docs to Your AI Assistant Select your tool to see how to add Weights & Biases Documentation's llms.txt as a documentation source. Cursor Windsurf Claude Code ChatGPT Zed Copilot 1. 1 Open any chat or composer panel 2. 2 Type @Docs and select "Add new doc" 3. 3 Paste the URL: https://docs.wandb.ai/llms.txt 4. 4 Wait for the green dot (indexing complete) 5. 5 Reference @Docs in chat when asking about Weights & Biases Documentation Cursor crawls the URL and indexes all subpages. Add a trailing slash to index child pages too. ## Spec Compliance 4 notes ## Frequently Asked Questions Where is Weights & Biases Documentation's llms.txt file? Weights & Biases Documentation publishes its llms.txt at https://docs.wandb.ai/llms.txt. This file provides a structured, markdown-formatted index of Weights & Biases Documentation's documentation that AI systems can consume to understand the project's APIs, guides, and references. How do I use Weights & Biases Documentation's llms.txt with AI coding assistants? Copy the llms.txt content and paste it into your AI assistant (Cursor, Windsurf, Claude, ChatGPT) as context. This gives the AI an accurate map of Weights & Biases Documentation's documentation so it can provide better code suggestions and answers about Weights & Biases Documentation. What does Weights & Biases Documentation's llms.txt contain? Weights & Biases Documentation's llms.txt contains 2 sections and 893 documentation links in 98.1 KB (~24.9k tokens). Key sections include Docs, OpenAPI Specs. How many tokens does Weights & Biases Documentation's llms.txt use? The concise llms.txt index is approximately 24.9k tokens (98.1 KB). Most AI assistants can fit this within their context window. For the full expanded documentation, look for an llms-full.txt variant which embeds all content inline. Can I generate an llms.txt for my own project? Yes. Use the free llms.txt generator at mdream.dev/tools/llms-txt/generator to crawl your site and produce a spec-compliant llms.txt file. You can also validate existing files with the llms.txt validator. ## Browse More llms.txt Files [DigitalOcean Documentation Comprehensive tutorials, references, example code, and more for DigitalOcean products.](https://mdream.dev/llms-txt/wandb/llms-txt/digitalocean) [Dagster Docs](https://mdream.dev/llms-txt/wandb/llms-txt/dagster) [ElevenLabs Documentation](https://mdream.dev/llms-txt/wandb/llms-txt/elevenlabs) [Hurl Documentation With `\[Query\]` section, params don't need to be URL escaped.](https://mdream.dev/llms-txt/wandb/llms-txt/hurl) [Unsloth Documentation Unsloth Studio is a web UI for training and running open models like Qwen, DeepSeek, gpt-oss and Gemma locally.](https://mdream.dev/llms-txt/wandb/llms-txt/unsloth) [Nylas | Powerful Communication APIs Unlock the power of your communications data. The Nylas communication APIs remove complexity & empower teams to create seamless user experiences.](https://mdream.dev/llms-txt/wandb/llms-txt/nylas-mail) [Fastify Fast and low overhead web framework, for Node.js](https://mdream.dev/llms-txt/wandb/llms-txt/fastify) [openmediavault](https://mdream.dev/llms-txt/wandb/llms-txt/openmediavault) [View all 99 projects ](https://mdream.dev/llms-txt/wandb/llms-txt) ## Related Tools [

llms.txt Generator

Generate an llms.txt for your own project.](https://mdream.dev/llms-txt/wandb/tools/llms-txt/generator) [

llms.txt Validator

Validate your llms.txt against the official spec.](https://mdream.dev/llms-txt/wandb/tools/llms-txt/validator) © 2026 [Harlan Wilton](https://github.com/harlan-zw) · [MIT](https://github.com/harlan-zw/mdream/blob/main/license) [GitHub](https://github.com/harlan-zw/mdream) [Discord](https://discord.com/invite/275MBUBvgP)