Ikibazo gihuza ibibazo · Hagati Urwego

Ikibazo-Abakinnyi Banegura Ikibazo

Kugenzura umukinnyi no kunegura inshingano, inyungu nibimenyetso bya TD, kugabanya itandukaniro, ibikorwa bikomeza hamwe nurugero rwa PPO. mubikorwa bigezweho.

Inzira zijyanye no kuyoboraUburyo bwo Kunegura Abakinnyi
Ikibazo 1 Bya 8

Muburyo bw'abakinnyi-banegura, ni uruhe ruhare 'kunegura'?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Muburyo bw'abakinnyi-banegura, ni uruhe ruhare 'kunegura'?
    • Kugereranya agaciro ka leta (cyangwa ibikorwa) no gusuzuma amahitamo yabakinnyi
    • Guhitamo igikorwa ugomba gukora
    • Kubika buffer isubiramo
    • Kugirango ushakishe ibidukikije
  2. Niki 'akarusho' A (s, a) = Q (s, a) - V (s) bipima iki?
    • Igihembo cyose
    • Nigute ibikorwa byiza biruta agaciro kagereranijwe ka leta
    • Igipimo cyo kwiga
    • Umubare wibikorwa urahari
  3. Ni ukubera iki kongeraho kunegura bigabanya itandukaniro ugereranije nuburyo bwiza bwa politiki-buhoro buhoro nka REINFORCE?
    • Ikuraho ibihembo muburinganire
    • Bituma politiki igenwa
    • Abanegura batanga ibyingenzi, ivugurura rero rikoresha ibimenyetso bitarangwamo urusaku
    • Yongera kugabanuka
  4. Ni ubuhe bushobozi uburyo Abakinnyi-banegura bafite ubwo buryo busanzwe bushingiye ku gaciro nka Q-Kwiga bitwaye nabi?
    • Ibikorwa byihariye
    • Gukoresha ibihembo
    • Kugaragaza ejo hazaza
    • Umwanya uhoraho wibikorwa
  5. Niki muri ibi bikurikira ni Umukinnyi-Critic-ishingiye kuri algorithm ikoreshwa cyane kuri RLHF kurugero rwindimi?
    • PPO (Gukwirakwiza Politiki Yegeranye)
    • K-bisobanura
    • Gusubira inyuma
    • Ikibazo
  6. Ni ikihe kimenyetso umukinnyi akoresha muguhindura politiki yacyo?
    • Iyinjiza rya pigiseli mbisi
    • Ibyiza cyangwa ikosa rya TD ryatanzwe nabanegura
    • Ingano yo gusubiramo
    • Umubare uteganijwe
  7. An action has a positive advantage estimate. How does a policy-gradient actor typically use that signal?
    • Decrease its probability in every state
    • Set the critic’s value estimate to zero
    • Increase the probability of choosing that action in the state
    • Replace the policy with a Q-table
  8. An action’s estimated advantage is negative. What direction should the actor update favor?
    • Increase it because the critic was surprised
    • Delete the state from the environment
    • Raise the reward discount factor automatically
    • Reduce the likelihood of that action relative to the policy baseline