quiz bu lëkkale ak tegtal · Diggante Niveau

Kontekst bu gudd ak Quiz RAG

Teg ci contexte bu gudd ak RAG ci njëg, latency, njub ak seddaay, boole ci effet bu réer ci digg bi ak jëmmal hybrid.

Yooni njiit yu jëm ciKontekst bu gudd ak RAG
Laaj 1 de 8

Lan moo waral approche long-context di gëna njëg laaj bu nekk RAG?

Xoolaatal laaj bu nekk8 laaj

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

  1. Lan moo waral approche long-context di gëna njëg laaj bu nekk RAG?
    • Providers charge per input token, and long context sends far more tokens
    • Long-context models require a vector database
    • RAG is always free
    • Long context requires fine-tuning for each question
  2. Lan la jàngat bu 2023 Lost in Middle gis?
    • Models ignore the first sentence of every prompt
    • Retrieval always beats long context
    • Performance dropped when relevant information was placed in the middle of long inputs
    • Context windows cannot exceed 4,000 tokens
  3. Lan mooy gàttal bu mag bi ci test yi ñuy def ci piccu ñax?
    • Ci nataal kese lañuy liggéey
    • They measure simple lookup, not reasoning across many scattered facts
    • They cannot be run on long inputs
    • They measure cost, not accuracy
  4. Ban situation moo gëna baax ci RAG ci contexte bu yàgg?
    • A single short policy document queried a few times
    • Summarizing the themes of one novel
    • Comparing two chapters of one report
    • A large, frequently changing corpus with per-user access permissions
  5. Sooy jëfandikoo cache prompt ak corpus bu gudd, naka lañu wara raññee prompt bi?
    • Put the question first and the corpus last
    • Shuffle the corpus on each request
    • Put the stable corpus first and the variable question last
    • Split the corpus across separate requests
  6. Why does time to first token increase with context length?
    • The model must process every input token before it starts answering
    • Longer prompts are sent over slower networks
    • The model retrieves documents from the internet
    • Output tokens are generated in parallel
  7. Which task is long context particularly well suited to?
    • Searching millions of support articles
    • Serving users with different clearance levels
    • Answering from data that changes every hour
    • Summarizing themes across an entire document
  8. What does a common hybrid of long context and RAG look like?
    • Using RAG only for images and long context only for text
    • Retrieving relevant documents generously, then including whole documents in a long window
    • Fine-tuning the model on the retrieved chunks
    • Alternating approaches on odd and even days