Guanyu-wang-EE

Use when RL/DRL/DL/ML paper reproduction or energy-system control work needs unattended long-goal execution, long training, full-resume checkpoints, mandatory TensorBoard long-run visualization when applicable, traceable train/test evaluation, same-tier raw environment reward delta comparisons when applicable, online-checkable CSV/JSONL/stdout records, IEEE TSG/TPS figures, Chinese-first colleague-readable reports, post-training PPT/report packs, figure/table quality audit, SHA256 manifests, artifact indexes, risk analysis, or cleanup/exclusion of failed or half-finished artifacts.

Use when planning, implementing, auditing, or reporting hyperparameter optimization for deep reinforcement learning algorithms such as PPO, SAC, TD3, DDPG, DQN, A2C, recurrent RL, or multi-agent RL.

Create compact, execution-ready multi-stage planning packages with one index, short stage plans, contract/spec files, scientific gates, stop/debug rules, and reproducibility handoffs. Use when designing or auditing staged project plans, long-goal prompts, RL/scientific experiment plans, review-response roadmaps, or any workflow that needs phase gates and evidence contracts.

Create, revise, split, or audit README.md files for Guanyu Wang's research/code repositories. Use whenever a task mentions README, readme, documentation index, GitHub rendering, bilingual English/Chinese README pairs, README_zh.md, artifact navigation, figure/table/report links, reproducibility handoff, Markdown math rendering, or README cleanup/push readiness.

Use when the user provides academic PDF files and wants EndNote .enw/.env import files, IEEE citations, DOI/arXiv/Crossref metadata checking, or systematic literature PDF renaming.