Kuiz berkaitan panduan · Sederhana Tahap

Mengelakkan AI Vendor Lock-In dengan Kuiz Strategi Berbilang Model

Menguji lapisan kunci masuk LLM, sebab menukar model pembenaman memaksa pengindeksan semula, dan cara suite penilaian dan penghalaan kanari menjadikan penukaran selamat.

Laluan panduan yang berkaitanAI Vendor Lock In Strategi Berbilang Model
soalan 1 daripada 8

Why does switching to a new embedding model usually mean re-embedding the whole corpus?

Pratonton setiap soalan8 soalan

Jawapan dan penjelasan muncul semasa anda mengambil kuiz.

  1. Why does switching to a new embedding model usually mean re-embedding the whole corpus?
    • New models refuse to read old databases
    • Vector databases delete vectors after a model change
    • Vectors from different embedding models are in different spaces and cannot be compared
    • Embeddings expire after a fixed number of days
  2. What is the main purpose of an evaluation suite in a multi-model strategy?
    • To measure whether a candidate model meets quality standards on your real tasks before switching
    • To get a lower price from the current vendor
    • To encrypt prompts before they are sent
    • To train a new model from scratch
  3. Why does putting every model behind one unified API not guarantee portability?
    • Unified APIs only work with open-source models
    • Unified APIs are always slower than direct calls
    • Unified APIs do not support streaming
    • Prompts tuned to one model's behavior often perform differently on another
  4. Sending simple tagging tasks to a small model and complex complaints to a frontier model is an example of what?
    • Penalaan halus
    • Routing tasks by difficulty to control cost
    • Zero data retention
    • Re-embedding
  5. What should be stored with every vector to make future re-indexing straightforward?
    • The user's password
    • The billing account number
    • The embedding model's name and version, along with source text and chunking settings
    • Only the vector dimensions
  6. Which of these is an example of platform-feature lock-in?
    • Relying on a provider's hosted conversation threads and built-in vector store
    • Sending requests in JSON
    • Using HTTPS for API calls
    • Logging response latency
  7. What is usually true of fine-tuned models on hosted provider platforms?
    • They can be downloaded and run anywhere
    • They automatically work with every other provider
    • They cost nothing to keep
    • They usually cannot be exported to another provider
  8. What does canary routing mean when adopting a new model?
    • Switching all traffic at once overnight
    • Sending a small share of traffic to the new model and comparing results before widening
    • Testing only with synthetic data
    • Letting users choose their model manually