Frozen Embedding

Definitions

1. An embedding layer or embedding representation whose parameters are kept fixed during some later stage of training or adaptation. Freezing embeddings can reduce training cost or preserve useful structure, though it may limit task-specific adaptation.

   In plain English: An embedding representation kept fixed instead of being updated during training.

   Example: "They started with a frozen embedding so the downstream classifier could train quickly without distorting the original semantic space."

Related Terms

• AI Vector
• Chinchilla Scaling
• Constitutional AI Method
• Context Distillation
• Embedding Dimension
• Embedding Layer
• Expert Iteration
• Goal Conditioned
• Gradient-Free Optimization
• Machine Unlearning
• Process Reward Model
• Replay Buffer