// Course · stage 5 of 9
Post-training, Alignment & Reasoning Models
SFT and PEFT, RLHF/DPO, RL with verifiable rewards, reasoning models and test-time compute, RL basics.
3 unitsabout 32 hours27 algorithms
01Units in this stage
In order. Each unit has goals, algorithms, must-read papers, a project and a self-test.
02Algorithms you will meet
Algorithms Atlas →Instruction Tuning (FLAN)Self-InstructAdaptersPrefix & Prompt TuningLoRA (Low-Rank Adaptation)QLoRADoRA (Weight-Decomposed LoRA)Model Merging & Task ArithmeticReward Modeling (Preference Learning)Reinforcement Learning from Human Feedback (RLHF)Direct Preference Optimization (DPO)KTO (Kahneman–Tversky Optimization)ORPO (Odds Ratio Preference Optimization)SimPO (Simple Preference Optimization)Constitutional AI / RLAIFChain-of-Thought PromptingSelf-ConsistencyTree of ThoughtsProcess Reward Models (PRM)STaR (Self-Taught Reasoner)Group Relative Policy Optimization (GRPO)RL with Verifiable Rewards (RLVR)DAPODeepSeek-R1 Reasoning RL RecipeTest-Time Compute ScalingBudget Forcing (s1)Self-Refine
Part of these routes: Application Developers · Algorithm Researchers · Product Managers & Leaders · Domain Scientists