← Skill graph
Policy gradient methods
Part of Reinforcement learning and RLHF
Needs first
Helps first
Nothing recommended beyond what is required.
Unlocks
Where it is assessed
7 assessed items, 0 done.
- AIE-206 M3.3 REINFORCE, from the derivation · tier 1
- AIE-206 M3.4 Variance reduction, measured · tier 2
- AIE-206 M3.6 Retention: policy gradients and variance · tier 1
- AIE-206 M4.4 Why DPO gets to skip all this · tier 2
- AIE-206 M5.1 rlhf-study, published · tier 3
- AIE-206 M5.2 Make it hack the reward · tier 2
- AIE-206 M5.4 Oral defense · tier 4