Wirewalk Instituteof AI
ven-160 Intermediate Waitlist

Meta Llama and Open-Weight Models

For engineers, architects and technical leads who want to deploy Meta's open-weight models with a clear understanding of the license, the hardware they need, and how Meta's model line has changed.

The curriculum is being prepared and reviewed. Enrollment is not open yet.

About this course

Llama made open-weight models a mainstream enterprise option, and many production systems still run Llama 3.x or Llama 4 checkpoints. By 2026 Meta's public model line has changed shape: Meta's developer site now leads with the Muse family, including Muse Spark Open, an open-weight model released in October 2026 under Meta's separate Muse Model License, and models offered through the Meta Model API, while the Llama 4 models remain available under their community license. This course gives technical teams a clear, current picture of what Meta publishes, under what terms, and what it takes to run it well.

You will read the Llama 4 Community License and Acceptable Use Policy clause by clause (attribution and naming rules, the large-platform user threshold, geographic restrictions on multimodal use) and see why 'open weight' is not the same as open source under the OSI definition. You will size GPU memory for total versus active mixture-of-experts parameters, download gated checkpoints from Hugging Face, serve a model with vLLM and llama.cpp, use the correct chat template and tool-calling format, add Llama Guard 4 and Prompt Guard 2 as safety classifiers, and evaluate a model on your own tasks before committing. The final module covers when to fine-tune, managed Llama on the major clouds, and how to pin versions and plan for roadmap changes. Facts are stated as of the review month with links to Meta's official documentation. Completion and a passing exam score earn a Wirewalk Institute of AI certificate of completion and assessment; it is not a Meta certification and is not endorsed by Meta.

What you will be able to do

  • Describe the evolution of Meta's model line through Llama 4 and the Muse family, and identify which models are open-weight and how each is accessed
  • Interpret the Llama 4 Community License, its acceptable use policy and the Muse Model License for a proposed deployment, including attribution, user or revenue thresholds and geographic limits
  • Estimate GPU memory and hardware for a dense or mixture-of-experts Llama model at a chosen precision and context length
  • Download, serve and smoke-test a Llama model with vLLM or llama.cpp using the correct chat template and tool-calling format
  • Add Llama Guard 4 and Prompt Guard 2 to an application and explain what each classifier does and does not catch
  • Evaluate a Llama model on a task-specific test set and decide between prompting, retrieval and fine-tuning
  • Plan a maintainable deployment with version pinning, managed-cloud alternatives and a response to upstream roadmap changes

Syllabus

5 modules · 15 lessons · about 12 hours

Module 1: Meta's Model Line in 2026

How Llama evolved, what Llama 4 changed, and where Meta's Muse family and Model API now fit.
3 lessons
  1. From Llama 2 to Llama 4: The Architecture Changes That Still Matter
  2. Llama 4 Scout and Maverick: Mixture-of-Experts, Native Multimodality and Long Context
  3. Muse Spark Open, Muse Glimmer and the Meta Model API: Where Meta's Open and Hosted Lines Split

Module 2: Licenses and Obligations

What Meta's license terms require, restrict and leave open, read as a deployment checklist.
3 lessons
  1. Reading the Llama 4 Community License: Attribution, Naming and the Large-Platform Threshold
  2. The Acceptable Use Policy and Geographic Limits on Multimodal Models
  3. Open Weight vs Open Source: The OSI Definition, the Muse Model License and Why Legal Teams Care

Module 3: Getting and Running the Weights

From gated download to a working, tested endpoint on hardware you have sized correctly.
3 lessons
  1. Gated Downloads from Hugging Face: Checkpoints, Formats and Integrity Checks
  2. Sizing Hardware for Total vs Active MoE Parameters, Precision and KV Cache
  3. Serving Llama with vLLM and llama.cpp: First Deployment and Smoke Tests

Module 4: Prompt Format, Tools and Safety

Talking to Llama correctly, wiring tools, and adding Meta's safety classifiers.
3 lessons
  1. Llama Chat Templates, System Prompts and Tool-Calling Format
  2. Llama Guard 4 and Prompt Guard 2: Content Safety and Prompt-Attack Classifiers
  3. Evaluating Llama on Your Own Tasks Before You Commit

Module 5: Adapting and Operating Llama

Fine-tuning decisions, managed hosting options and keeping a deployment maintainable.
3 lessons
  1. When to Fine-Tune Llama Instead of Prompting or Adding Retrieval
  2. Managed Llama and Muse on Amazon Bedrock, Microsoft Foundry, Google Cloud Model Garden and Specialist Hosts
  3. Version Pinning, Security Patching and Planning for Meta's Roadmap Changes

Exam and certificate

The exam unlocks once you have completed every lesson. It draws 20 questions from a larger bank, spread across the modules, with the answer options shuffled for each attempt. You have 40 minutes and need 80% to pass. After a failed attempt you can try again after 24 hours, up to 3 attempts in total. Feedback is given per module, never as the answer key.

Passing earns a certificate of completion and assessment from the Wirewalk Institute of AI with a unique ID that anyone can check at institute.wirewalk.com/verify. It is not an accredited qualification or a vendor certification. How certification works.

The Wirewalk Institute of AI is independent. This course teaches publicly documented products and is not affiliated with or endorsed by Meta (Llama, Muse), Hugging Face, vLLM, llama.cpp, Amazon (Bedrock), Microsoft (Foundry), or Google Cloud (Model Garden). Product and company names are trademarks of their respective owners.

More in AI Models and Vendors

All 22 courses in this track
ven-001 Free intro Available

Your First Hour with ChatGPT, Claude and Gemini

A hands-on hour with ChatGPT, Claude and Gemini, told through one small business racing a deadline. Set up an account with a privacy setting you chose, decide what never goes into a personal account, write prompts that produce usable drafts, use files and web search, catch confident mistakes before they go out, and pick the assistant that fits how you work.

1 h · 7 lessons Free intro
ven-110 Basic Available

ChatGPT and the OpenAI Platform for Professionals

A practical, current tour of ChatGPT as a professional tool: choosing a plan and model, prompting well, analyzing files, researching with citations, using images and voice, organizing work in projects and memory, connecting apps and plugins, delegating multi-step work to ChatGPT Work, and knowing which admin and privacy controls matter. Ends with clear guidance on moving a workflow to the OpenAI API.

8 h · 15 lessons See price at checkout
ven-120 Basic Available

Claude for Professionals

Learn Claude by solving one firm's real problem: getting through renewal season with too few analysts and without putting client data in the wrong place. Each module compares the approaches professionals actually choose between, shows a realistic failure and its fix, and ends with a Decision Lab where you make the call under pressure.

12 h · 15 lessons See price at checkout