Untitled

Large Language Model (LLM)

An artificial intelligence model designed to understand and generate human-like text based on vast amounts of training data, typically consisting of billions or trillions of parameters.

Encoder

Part of LLM architectures that encode text into embeddings, converting sequences of words into vectors to capture text semantics or meaning.

Decoder

Component of LLM architectures that decodes or generates text, producing a sequence of words based on input sequences.

Prompting

Providing an initial input or "prompt" to LLMs to guide text generation or task completion based on training data.

In-context learning

Constructing prompts with task demonstrations for LLMs.

Greedy decoding

Decoding technique where the model selects the token with the highest probability at each step of sequence generation.

Retrieval Augmented Generation (RAG)

Approach combining retrieval-based and generative models to enhance text generation with relevant external knowledge.

Vector Databases

Specialized databases for storing and querying high-dimensional vector data, suitable for ML, NLP, and computer vision applications.

Semantic Search

Advanced information retrieval technique focusing on understanding the meaning behind search queries and documents for more accurate results.

OCI Generative AI Service

Oracle Cloud service offering pre-trained LLMs, fine-tuning capabilities, dedicated resources, and flexibility for text-based tasks.