AI Dictionary › AI Fundamentals

Encoder-Decoder

The encoder-decoder architecture is a two-part scheme designed to transform one sequence into another, such as translating a sentence between languages. The encoder reads the entire input and compresses it into a dense internal representation, a numerical summary; the decoder starts from that representation and generates the output piece by piece. It's like an interpreter who first listens to and deeply understands a whole speech, forming a mental idea, then restates it in the target language. This separation lets input and output have different lengths and structures.

Definition

The scheme underpins machine translation, summarization, and speech synthesis, and inspired architectures like the transformer. Distinguishing it from decoder-only models, used by many generative chatbots, clarifies why some systems excel at transforming given text while others are optimized to freely generate new content.

Related terms

• Transformer
• Attention Mechanism
• Positional Encoding
• Diffusion Model
• Autoencoder

More in AI Fundamentals

• LLM (Large Language Model)
• Token
• Context Window
• Temperature
• AI Hallucination
• Fine-tuning
• RAG (Retrieval Augmented Generation)
• Embedding

Put it into practice

• Real scenario: apply this concept with an instant AI verdict.
• Daily gym: short drills to train dictionary concepts.
• Playoff game: challenge other players, elimination-style.

📱 Download the Android app (beta) iOS coming soon

Say what you mean. Get what you need.

Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.