Skip to content
AImpact
IT EN
Models Advanced Also known as: Rotary Position Embedding · Embedding posizionale rotatorio

RoPE

/rope/

A positional encoding technique that rotates token vectors based on their position, baking order directly into attention.

ShareLinkedInX

In practice

It has become the de facto standard: Llama, Mistral, Qwen, DeepSeek, and GPT-4 class models all use it. It lets you stretch the context beyond the training length with tricks like NTK-aware or YaRN. If you fine-tune on long contexts, understanding RoPE is almost mandatory.

Related terms

Positional Encoding Transformer Attention Context window

Seen in the wild

10 entries mentioning it
  1. EU AI Act: 100-day countdown to the high-risk system rules
    High
  2. Mistral relaunches with a new open-weight reasoning flagship
    Medium
  3. EU AI Act: General-Purpose AI rules enter into force
    High
  4. Google A2A Protocol: open standard for communication between heterogeneous AI agents
    High
  5. Google ADK + A2A: open-source framework and protocol for agents that talk to each other
    High
  6. FlashInfer 0.2: attention library for LLM serving with paged KV cache and RoPE fusion
    Medium
  7. GGUF specification: the standard format for local quantized LLM models
    Medium
  8. EU AI Act: European Parliament adopts the first comprehensive AI law
    Landmark
  9. Mistral Large and Le Chat: Mistral's commercial pivot with Microsoft partnership
    Medium
  10. Mistral 7B: Europe joins the open-source race
    High
← All terms