Pith. sign in

Simple statistical gradient-following algorithms for connectionist reinforcement learning

2 Pith papers cite this work. Polarity classification is still indexing.

2 Pith papers citing it

fields

cs.LG 2

years

2026 2

verdicts

UNVERDICTED 2

representative citing papers

OGPO: Sample Efficient Full-Finetuning of Generative Control Policies

cs.LG · 2026-05-04 · unverdicted · novelty 6.0 · 2 refs

OGPO enables sample-efficient full-finetuning of generative control policies via off-policy critics and modified PPO, achieving SOTA on robot manipulation tasks while rescuing poorly initialized behavior cloning policies without expert data.

citing papers explorer

Showing 2 of 2 citing papers.

  • OGPO: Sample Efficient Full-Finetuning of Generative Control Policies cs.LG · 2026-05-04 · unverdicted · none · ref 50 · 2 links

    OGPO enables sample-efficient full-finetuning of generative control policies via off-policy critics and modified PPO, achieving SOTA on robot manipulation tasks while rescuing poorly initialized behavior cloning policies without expert data.

  • Beyond Distribution Sharpening: The Importance of Task Rewards cs.LG · 2026-04-17 · unverdicted · none · ref 42

    Task-reward reinforcement learning yields robust gains on math benchmarks for models like Llama-3.2-3B while distribution sharpening alone delivers only limited and unstable improvements.