AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

HyMTRL: A Hybrid Multi-Task Reinforcement Learning Framework via Phased Policy Evolution

Jinmin He, Kai Li, Xiaoyi Dong, Yifan Zang, Yuheng Jing, Yifan Zhang, Junliang Xing, Jian Cheng

ICML 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Off-policy actor-critic with expressive policies

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add HyMTRLMOORE · MetaWorld-MT50success rate (%), all 50 tasks: 68.1 -> 86.2Improved
add HyMTRLMTSAC · MetaWorld-MT50success rate (%), all 50 tasks: 48.4 -> 71.7Improved
add HyMTRLSM · MetaWorld-MT50success rate (%), all 50 tasks: 55.6 -> 70.4Improved
add HyMTRLMOORE · MetaWorld-MT50bottom-10 success rate (%): 0.8 -> 37.3Improved
add HyMTRLMHSAC · MetaWorld-MT50bottom-10 success rate (%): 0.0 -> 0.0No clear effect
add HyMTRLMHSAC · MetaWorld-MT10success rate (%): 62.3 -> 80.1Improved
add critic reset (critic only) (ablation)HyMTRL MHSAC · MetaWorld-MT50success rate (%), all 50 tasks: 65.9 -> 72.6Improved
reset actor as well as critic (ablation)HyMTRL MHSAC · MetaWorld-MT50success rate (%), all 50 tasks: 72.6 -> 67.9Hurt
add critic reset to baseline without HyMTRL (ablation)MHSAC · MetaWorld-MT50success rate (%), all 50 tasks: 51.6 -> 57.1Improved
raise mastery threshold from 0.9 to 0.99 (ablation)HyMTRL MHSAC · MetaWorld-MT50success rate (%), all 50 tasks: 72.6 -> 59.4Hurt
lower mastery threshold from 0.9 to 0.5 (ablation)HyMTRL MHSAC · MetaWorld-MT50success rate (%), all 50 tasks: 72.6 -> 60.7Hurt
use critic reset interval 3e5 vs 1e5 (ablation)HyMTRL MHSAC · MetaWorld-MT50success rate (%), all 50 tasks: 53.7 -> 72.6Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy