← All notes

Summary

Proximal Policy Optimization Algorithms (2017) introduced PPO, a policy-gradient algorithm that is still the workhorse behind much of RLHF and later reasoning-model training. It aimed to get much of the reliability of earlier trust-region methods with a simpler, more practical update.

What it introduced: the PPO clipped (and related) objectives for updating a policy without taking steps that are too large, along with a practical training algorithm that became a default in deep RL.

What it led to: the usual optimizer in early RLHF pipelines, including InstructGPT, and a baseline later papers still compare against even when they replace PPO with other preference methods. Live notes: 2204-05862-helpful-harmless-rlhf and 2305-18290-direct-preference-optimization. See also this batch's pointer to deep RL from human preferences.

Where to read it: https://arxiv.org/abs/1707.06347 (v2, last revised 28 Aug 2017).

The paper is posted on arXiv under the arXiv non-exclusive distribution license. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: arXiv non-exclusive distribution; this note links to it and does not reproduce it.

@misc{iil_note_jh6jnn28,
  author = {{'iinventedlight (AI-assisted)'}},
  title = {{Proximal Policy Optimization Algorithms}},
  howpublished = {iinventedlight Library},
  year = {2026},
  date = {2026-11-11},
  url = {https://iinventedlight.com/library/notes/1707-06347-ppo/},
  note = {Permanent ID iil:note:jh6jnn28. License: CC-BY-4.0.},
}

Markdown

Spotted a mistake? Use the MCP feedback tool or email hello@iinventedlight.com.

For language models and agents: start at /llms.txt | /agents.md | /.well-known/agent-library.json. This page in Markdown: /library/notes/1707-06347-ppo.md.