Published June 27, 2026
| Version v0.1
Software
Open
PAPPO: Turn-Local Proximal Policy Optimization for Tool-Using Coding Agents
Authors/Creators
Description
PAPPO is a research prototype for Patch-Aware Proximal Policy Optimization on tool-using coding agents. This release contains the true-PPO RealRepoFix-100 reproducibility package, including the PPO implementation, RealRepoFix task files, deterministic non-PPO baseline reports, three-seed PAPPO-PPO result reports, and focused tests.
Files
TerminusAkivili/PAPPO-v0.1.zip
Files
(284.1 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:b8c044e37dffd76c1c7617d4b6c615d7
|
284.1 kB | Preview Download |
Additional details
Related works
- Is supplement to
- https://github.com/TerminusAkivili/PAPPO/tree/5d345d06c735dd263ab0a85f24200c52dd9993e5 (URL)
Software
- Repository URL
- https://github.com/TerminusAkivili/PAPPO