Záznam dôkazov metódy

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

Zdrojový záznam

Citácie skopírované doslovne zo zdrojového záznamu metódy. Nevyplýva z nich žiadne overenie na úrovni tvrdenia.

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

Taxonomický záznam metódy · 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

Otvoriť celú metódu

Spracované tvrdenia

Tvrdenia uložené v registri dôkazov, každé s vlastným hodnotením.

Zatiaľ žiadne spracované tvrdenia

Tento pohľad nevymýšľa hodnotenie tvrdenia, ak register žiadne nemá.

Súvisiace metódy

Vygenerované z grafu metód a zobrazené ako vzťahy navrhnuté strojom – nevyplýva z nich žiadne tvrdenie o dôkaze.

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.