FineTuningMethodType

Enums · AiDotNet.Interfaces

Specific fine-tuning method types.

How It Works

This enum provides a complete list of supported fine-tuning methods based on the latest research (2023-2025).

Fields

Field Summary
FineTuningMethodType APO Adversarial/Anchored Preference Optimization.
FineTuningMethodType CPO Contrastive Preference Optimization.
FineTuningMethodType CalDPO Calibrated DPO.
FineTuningMethodType ConstitutionalAI Constitutional AI - principle-based alignment.
FineTuningMethodType DPO Direct Preference Optimization.
FineTuningMethodType GAPO Generative Adversarial Policy Optimization.
FineTuningMethodType GRPO Group Relative Policy Optimization (DeepSeek).
FineTuningMethodType IPO Identity Preference Optimization - addresses DPO overfitting.
FineTuningMethodType KTO Kahneman-Tversky Optimization - uses prospect theory.
FineTuningMethodType KnowledgeDistillation Knowledge Distillation from teacher model.
FineTuningMethodType LCPO Latent Collective Preference Optimization.
FineTuningMethodType NCA Noise Contrastive Alignment.
FineTuningMethodType ORPO Odds Ratio Preference Optimization - combines SFT and alignment.
FineTuningMethodType PPO Proximal Policy Optimization.
FineTuningMethodType PRO Preference Ranking Optimization.
FineTuningMethodType RDPO Robust DPO.
FineTuningMethodType REINFORCE REINFORCE algorithm.
FineTuningMethodType RLAIF Reinforcement Learning from AI Feedback.
FineTuningMethodType RLHF Reinforcement Learning from Human Feedback with PPO.
FineTuningMethodType RLVR Reinforcement Learning with Verifiable Rewards.
FineTuningMethodType RRHF Rank Responses to align Human Feedback.
FineTuningMethodType RSO Rejection Sampling Optimization.
FineTuningMethodType SFT Supervised Fine-Tuning - standard labeled data training.
FineTuningMethodType SLiCHF Sequence Likelihood Calibration with Human Feedback.
FineTuningMethodType SPIN Self-Play Fine-Tuning.
FineTuningMethodType SafeEXO Safe Externalized Optimization.
FineTuningMethodType SafeNCA Safe Noise Contrastive Alignment.
FineTuningMethodType SimPO Simple Preference Optimization - reference-free.
FineTuningMethodType TPO Triple Preference Optimization.
int value__