Ikibazo gihuza ibibazo · Hagati Urwego

MLflow na Model Lifecycle Ikurikirana Ikibazo

Witoze MLflow igerageza ikurikirana, gupakira icyitegererezo, hamwe na verisiyo yerekana, harimo uruhare rwa aliase mubikorwa byoherejwe.

Inzira zijyanye no kuyoboraMlflow Nicyitegererezo Cyubuzima
Ikibazo 1 Bya 8

Ni ikihe kibazo cyibanze MLflow Tracking ikemura kubashakashatsi ba data?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Ni ikihe kibazo cyibanze MLflow Tracking ikemura kubashakashatsi ba data?
    • Mu buryo bwikora kuranga amakuru yimyitozo mbisi
    • Shishoza uburemere bwikitegererezo kuruhuka
    • Kubika inyandiko igereranya y'ibipimo, ibipimo, n'ibihangano mubigeragezo byinshi
    • Gusimbuza ibikenewe bya Kubernetes
  2. Componenti ya MLflow ibika versions z'inyandiko z'inyandiko n'ibyo tags cyangwa aliases?
    • Icyitegererezo cy'icyitegererezo
    • The learning-rate scheduler
    • The training-data labeler
    • The Kubernetes pod scheduler
  3. Bisobanura iki ko MLflow ari 'framework-agnostic'?
    • Ikorana na TensorFlow gusa
    • Birasaba kwandika moderi mururimi rwihariye
    • Ikora kuri Databricks gusa
    • Ikora mumasomero menshi ya ML nka scikit-wige, PyTorch, na XGBoost
  4. Niyihe ntego yimiterere ya MLflow Model?
    • Guhagarika amakuru yamahugurwa
    • Gupfundikira icyitegererezo cyamenyerejwe hamwe nuburyohe bwa metadata kugirango gishobore koherezwa muburyo bwinshi
    • Guteganya kongera imyitozo
    • Kugereranya amashusho y'urusobe rw'imitsi
  5. Niyihe sosiyete yashizeho MLflow?
    • Imibare
    • Google
    • Netflix
    • Amazon
  6. Nigute MLflow yaguka kugirango ishyigikire AI ibyara inyungu?
    • Muguhagarika inkunga kubintu bya kera bya ML
    • Muguhinduka ibicuruzwa bifunze-isoko
    • Mugusimbuza ibyiciro byose byamahugurwa
    • Wongeyeho gushakisha, gucunga byihuse, no gusuzuma porogaramu za LLM
  7. A deployment resolves an MLflow model through a champion alias. What can reassignment of that alias change?
    • The recorded metrics inside every old experiment
    • The registered model version that the named reference resolves to
    • The model framework of all stored artifacts
    • The training labels in a historical dataset
  8. Two MLflow runs use different learning rates. What record helps compare their validation results?
    • Only the wall color of the training room
    • A registry alias without any run information
    • Each run’s logged parameters and validation metrics
    • A notebook filename without the experiment outputs