Meetodi tõendite kirje

Direct Preference Optimization

Direct Preference Optimization (DPO) is a training method introduced by Rafailov et al. in 2023 that aligns language models with human preferences without requiring an explicit reward model. By directly optimizing for preference pairs (better response vs worse response), DPO simplifies the training pipeline compared to reinforcement learning from human feedback (RLHF).

Sources recorded, not reviewed

Allikakirje

Tsiteeringud kopeeritud meetodi allikakirjest sõna-sõnalt. Nendest ei saa järeldada väidete tasemel kinnitust.

Direct Preference Optimization: Your Language Model is Secretly a Reward Model

Taksonoomiline meetodikirje · ml-model / deep-learning

Rafailov, R., Sharma, A., Mitchell, E., Manning, C. D., Ermon, S., & Finn, C. (2023). Direct preference optimization: Your language model is secretly a reward model. arXiv preprint arXiv:2305.18290. · URL

Ava täielik meetod

Kureeritud väited

Väited on salvestatud tõendite registrisse, igal oma hinnanguga.

Kureeritud väiteid veel pole

See vaade ei loo väite hinnangut, kui registris seda pole.

Seotud meetodid

Genereeritud meetodigraafist ja kuvatud masina soovitatud seostena – väiteid ei järeldata.

Same method familyLatent Diffusion Modelsmachine-suggested · Relational suggestion, not evidence.Same method familyMamba (State Space Model)machine-suggested · Relational suggestion, not evidence.Same method familyMasked Autoencodersmachine-suggested · Relational suggestion, not evidence.Same method familyQLoRAmachine-suggested · Relational suggestion, not evidence.