Awesome_life_long_rl_2025

Fine-tuning Reinforcement Learning Models is Secretly a Forgetting Mitigation Problem https://arxiv.org/pdf/2402.02868v3 Lifelong Reinforcement Learning with Modulating Masks https://arxiv.org/pdf/2212.11110 This has some connection with LLM + adapter Policy model.

March 16, 2025 · 1 min · 28 words · Sukai Huang

Awesome LLMs with Different Abstraction of Language Data 2025

Overview there is a lack of this research on how different level of abstraction / granularity of language instructions could affect LLMs performance in instruction following or other works. ABSINSTRUCT: Eliciting Abstraction Ability from LLMs through Explanation Tuning with Plausibility Estimation https://arxiv.org/pdf/2402.10646 Inference Helps PLMs’ Conceptual Understanding: Improving the Abstract Inference Ability with Hierarchical Conceptual Entailment Graphs https://aclanthology.org/2024.emnlp-main.1233.pdf That is the end!

March 16, 2025 · 1 min · 62 words · Sukai Huang

Survey of LLMs for Planning 2025

PlanGenLLMs: A Modern Survey of LLM Planning Capabilities https://arxiv.org/pdf/2502.11221 Understanding the planning of LLM agents: A survey https://arxiv.org/pdf/2402.02716 LLMs as Planning Modelers: A Survey for Leveraging large Language Models to Construct Automated Planning Models https://openreview.net/pdf?id=ebJIJkQjcE A Survey on Large Language Models for Automated Planning https://arxiv.org/pdf/2502.12435

March 13, 2025 · 1 min · 45 words · Sukai Huang

HTN planning @ Pascal Bercher ANU

Overview of what is hierarchical planning 1. Daniel Hoeller’s dissertation https://oparu.uni-ulm.de/items/a6c64b47-76e7-4532-8179-3e215a9eac9c It has a summary of what is hierarchical planning comment: a $t$ is a task id that can either refer to a $c \in C$ or an $a \in A$, but a method decompose $c$ only. Ok I see, the point, the whole thing trys to allow the set to have duplicate (? why not just claim that you have a multiset) ...

March 5, 2025 · 3 min · 451 words · Sukai Huang

Learning General Policies Through Sketch @ Hector Geffner

I will list some important literatures about the topic of learning general policies through sketches The research is initiated by Blai Bonet and Hector Geffner The high level goal of the research is as follows [!IMPORTANT] The construction of reusable knowledge (transfer learning) is a central concern in (deep) reinforcement learning, but the semantic and conceptual gap between the low level techniques that are used, and the high-level representations that are required, is too large. ...

March 4, 2025 · 3 min · 555 words · Sukai Huang