# Apple — REVERSAL-BENCH: A Reversibility Axis and Reset Oracle for Measuring the Reset-Free RL Cliff

- Company: Apple (apple.com)
- Announced: 2026-09-17
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://machinelearning.apple.com/research/reversal-bench-rl-cliff
- Record: https://forck.live/items/11728-reversal-bench-a-reversibility-axis-and-reset-oracle-for-measuring-the-reset
- Subject: Machine Learning Research

Apple introduces REVERSAL-BENCH, a benchmark that controls environmental reversibility via a continuous parameter and provides a reset oracle to test state recoverability across eight manipulation settings in five physics engines. Evaluating various policy architectures reveals a sharp reversibility cliff: reset-free agents are consistently absorbed into irrecoverable states as reversibility increases, whereas episodic agents maintain steady learning. The benchmark suite, a large multi-simulator dataset labeled with recoverability, and a reset oracle are released.

## Evidence

Verbatim from https://machinelearning.apple.com/research/reversal-bench-rl-cliff:

> We introduce REVERSAL-BENCH, a benchmark that controls reversibility via a continuous parameter ρ∈ [0, 1] and provides a reset oracle, a ground-truth verification mechanism to test state recoverability across eight manipulation settings in five physics engines.

---

Record: https://forck.live/items/11728-reversal-bench-a-reversibility-axis-and-reset-oracle-for-measuring-the-reset
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md