Llama
Llama is a family of large language models developed by Meta.
LLMs in Production
LLaMA is a family of autoregressive large language models released by Meta for research and development.
- How it is used
- Supports LLM experimentation, data-driven model development, fine-tuning, quantized serving, and production deployment.
- How deep it goes
- Descends from model variants and memory estimates into a simplified architecture and training workflow, while deferring full research replication.
- Examples
- Uses Llama 2 and Llama 3 variants, GPU sizing calculations, a SimpleLlama PyTorch model, training logs, and generated text.
- What you can do after
- Build it from scratch, train a simplified model, adapt it with LoRA, quantize it, and deploy it to a shareable production endpoint.
Natural Language Processing in Action, Second Edition
Llama 2 is a family of autoregressive large language models released by Meta for text generation and understanding.
- How it is used
- Deploys an open model for local text generation while illustrating the costs and limits of scaling language models.
- How deep it goes
- Explains model sizes, resource requirements, authentication, tokenization, generation, and streaming, but does not cover training internals.
- Examples
- Uses question-answer prompts, arithmetic reasoning, decoded token sequences, and a streaming generation loop.
- What you can do after
- Apply Llama 2 via Hugging Face Transformers to tokenize prompts, generate responses, decode output, and inspect reliability.
Read how Natural Language Processing in Action, Second Edition explains Llama →
Starting Data Analytics with Generative AI and Python
A family of large language models developed by Meta for generative-AI applications.
- How it is used
- Supports generative-AI applications through hosted API calls or local Python workflows.
- How deep it goes
- Explains API configuration and local execution steps, but defers detailed model internals and current library requirements.
- Examples
- Demonstrates hosted client configuration and local prompt invocation with a quantized GGUF model.
- What you can do after
- Build it from scratch by configuring a hosted Llama endpoint or invoking a downloaded model locally with Python.
Read how Starting Data Analytics with Generative AI and Python explains Llama →
Software Testing with Generative AI
Llama-2 is an open-source family of large language models released by Meta that can be downloaded and further trained.
- How it is used
- Positions Llama as an accessible foundational model for experimentation, adaptation, and AI community development.
- How deep it goes
- Defines its openness, sizes, and access trade-offs but does not detail model architecture, training procedures, or evaluation.
- Examples
- Uses 8-billion- and 70-billion-parameter variants and the need for self-sourced datasets and infrastructure as concrete examples.
- What you can do after
- Evaluate/understand Llama’s openness, model-size options, and practical requirements for training or deployment.
Read how Software Testing with Generative AI explains Llama →
Build Applications with Local AI Models on a Mac
A family of large language models developed by Meta.
- How it is used
- Provides a general-purpose starting point when comparing and selecting language models.
- How deep it goes
- Summarizes releases and capabilities without explaining model implementation.
- Examples
- Lists Llama 4 Scout and Maverick alongside earlier Llama releases.
- What you can do after
- Evaluate Llama as a model choice using size, speed, quality, and task suitability.
Read how Build Applications with Local AI Models on a Mac explains Llama →
Also covered in
- AI-Powered Developer: Build software with ChatGPT and Copilot
- Time Series Forecasting Using Foundation Models
- Transformers in Action
- Knowledge Graphs and LLMs in Action
- Introduction to Generative AI, Second Edition
- Quantization and Fast InferenceIn early access (MEAP)
What to know first
- language modelA language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- fine-tuningFine-tuning is the continued training of a pretrained model on targeted data or objectives to adapt its parameters or behavior to a task, domain, or use.
- GPT-4GPT-4 is a multimodal large language model developed by OpenAI that processes text and image inputs and generates text.
- logitA logit is a real-valued quantity representing log-odds for a binary outcome or an unnormalized model score for a candidate outcome before probability normalization.
Related concepts
- open-weight modelkind of · An open-weight model is a machine-learning model whose trained parameter values are made available for others to use, subject to applicable terms, without necessarily making its training data or source code available.
- language modelkind of · A language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- open-source modelkind of · An open-source model is a machine-learning model whose relevant artifacts, such as code or weights, are made available under terms permitting specified forms of access, use, modification, or redistribution.
- Ollamaused by · Ollama is software for downloading, managing, running, and serving language models locally.
- anomaly detectionuses · Anomaly detection is the identification of observations, events, or behaviors that substantially deviate from expected patterns or a normal baseline.
- api calluses · An API call is a programmatic request or invocation sent through an application programming interface to retrieve data or invoke an exposed operation.
Questions about Llama
What is Llama?
Llama is a family of large language models developed by Meta.
Which Manning books explain Llama?
LLMs in Production, Natural Language Processing in Action, Second Edition, Starting Data Analytics with Generative AI and Python, Software Testing with Generative AI and Build Applications with Local AI Models on a Mac.
What should I know before learning Llama?
Manning books that teach Llama usually introduce language model, fine-tuning, GPT-4 and logit first.
What is Llama a kind of?
Llama is a kind of open-weight model. An open-weight model is a machine-learning model whose trained parameter values are made available for others to use, subject to applicable terms, without necessarily making its training data or source code available.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.




