Codex

As explained in Generative AI in Action

Codex is a language model trained particularly on source code to generate and assist with programming code.

Codex is a language model trained on natural language and billions of lines of source code to generate and assist with programming. It powered the original GitHub Copilot and supports multiple programming languages and code-related tasks. section 5.1.3 section 2.1

Codex supports programming assistance through:

• Generating code from natural-language prompts and surrounding code context. section 5.1.3
• Supporting languages including C#, Java, Python, JavaScript, SQL, Go, PHP, and Shell. section 2.1
• Enabling completion workflows that can fill missing code between existing content through fill-in-the-middle generation. section 5.1.3
• Helping with legacy-code understanding and unit-test generation through richer developer interactions. section 5.1.3

Figure 5.7 Copilot completion lifecycle

The completion lifecycle shows how Copilot combines prompts, the active code file, open tabs, and broader solution context before generating suggestions. section 5.1.3

Codex is now deprecated and retired; its learnings have been incorporated into mainline GPT models, while newer Copilot Chat versions use newer models. section 5.1.3

How Generative AI in Action covers it

How deep it goes

Explains training, language coverage, contextual prompting, and fill-in-the-middle completion, but stops short of model internals.

Examples

Uses Copilot completions, contextual code suggestions, legacy-code assistance, and unit-test generation as concrete applications.

What it assumes

Assumes familiarity with programming code and developer workflows while introducing Codex as a code-focused language model.

What you can do after

Apply it via GitHub Copilot to generate code and complete programming tasks across multiple languages.

How it connects

Connects Codex to foundational models, GPT-3, GitHub Copilot, Copilot Chat, and fill-in-the-middle generation.

liveBookmanning.com

Read Generative AI in Action on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Introduction to Generative AI, Second EditionAn OpenAI GPT-based language model fine-tuned to generate and understand computer code.

Compare how 2 books explain Codex →

Related concepts

• GitHub Copilotused by · GitHub Copilot is an AI-powered programming assistant that provides code suggestions, generation, and other software-development assistance based on prompts and programming context.
• GPT modelkind of · A GPT model is a generative pretrained Transformer language model that produces text autoregressively by predicting successive tokens from preceding context.
• language modelkind of · A language model is a computational model that represents patterns in linguistic tokens to estimate sequence probabilities and predict or generate language from context.
• fine-tuningconfigured by · Fine-tuning is adapting a pretrained machine-learning model by training it further on targeted data or objectives for a task, domain, or behavior.
• Copilotimplements · Copilot is a Microsoft-branded family of AI assistant products that provide task-oriented assistance within software and services, including conversational, content-generation, and coding capabilities.
• conversational aiused by · Conversational AI is artificial intelligence that enables software systems to interact with users through natural-language dialogue.

This explainer was generated by Manning's knowledge graph from the text of Generative AI in Action. The section links point to the parts of the book it draws on.