AI

OpenAI

• Cloud computing: primarily Azure
• Frameworks: PyTorch

Anthropic

• Cloud computing: AWS, GCP
• Frameworks: PyTorch, JAX, and Triton

AI Models vs AI Agents

• language models can't take actions - they just output text.
• Agents are systems that use LLMs as reasoning engines to determine which actions to take and the inputs necessary to perform the action.

agent development frameworks: CrewAI and LangChain, Google Agent Development Kit (ADK).

MCP vs A2A

• take actions: tool-calling or function-calling, MCP
• collaborate with other agent: Agnet2Agent(A2A)

RAG vs CAG

RAG (Retrieval-Augmented Generation) and CAG (Cache-Augmented Generation) are two approaches for integrating external knowledge into large language models (LLMs), but CAG preloads data for faster, more efficient responses, while RAG dynamically retrieves information.

What is a checkpoint?

In the context of Large Language Models (LLMs), a checkpoint is a saved snapshot of the model's state at a specific point in time during training, allowing for efficient recovery and fine-tuning.

AI for coding

• gemini (Google) + cline (Open Source) in vscode https://cline.bot/
• copilot (Microsoft) + openai or other models in vscode

copilot vs cursor

• copilot is a vscode plugin
• cursor is a vscode fork

Cline vs Roo

• Roo Code is a fork of Cline