54 skills · sorted by relevance
Compile and improve existing English or Chinese academic papers written in Typst. Check references, formatting, prose, argument flow, abstracts, tables, and pseudocode, with source-level review comments and support for adapting papers to another publication venue.
Create figures for machine learning papers from text or experiment data. Uses Gemini for architecture diagrams and matplotlib/seaborn for numerical charts, choosing suitable layouts and chart types while preparing consistent styling, export files, and reusable generation scripts.
Polishes academic English paragraph by paragraph while preserving the author's meaning and technical content. Reviews grammar, word choice, voice, logical flow, and sentence structure, providing specific revision reasons alongside complete polished paragraphs.
Prepare LaTeX manuscripts for submission by applying publication templates, managing references, fixing compilation problems, and setting up automated PDF builds. The workflow also covers formatting checks and packaging submission files into a zip archive.
Use this skill whenever the user wants to generate detailed English prompts for AI image tools (NanoBanana / Gemini / DALL-E / Midjourney) to produce top-conference-quality academic figures — including framework diagrams, network architecture diagrams, pipeline flowcharts, module detail diagrams, comparison/ablation figures, and data pattern grids — especially when the user says "论文配图提示词", "生成论文配图", "学术论文生图", "架构图提示词", "框架图提示词", "顶会风格配图", "CVPR 风格图", "NeurIPS 风格图", "paper figure prompt", "academic diagram prompt", or provides a LaTeX/PDF/Word paper and asks for figure prompts. If the user has not specified a color scheme, present the 8 preset palette options and color tool links before generating any prompt.
The English summary is being prepared. Showing the original description.
Generate prompts for AI image tools to produce modern ML/RL paper-style figures matching the aesthetic of recent ICLR/NeurIPS/ICML 2024-2025 publications. Key traits — pure white canvas, white panels with soft drop shadow, rounded friendly font (Nunito/Poppins), densely packed but uncluttered floating elements, small pastel token squares, rich inline illustrations, colored keyword text, pill-shaped concept labels. Trigger phrases — "pastel风格论文配图", "现代ML论文配图", "modern ML figure prompt".
The English summary is being prepared. Showing the original description.
Standards and audits for writing academic papers, especially ML/NLP submissions to NeurIPS, ICML, ICLR, ACL, EMNLP, NAACL. Covers the author's house prose rules (zero em dashes, zero fluff, every sentence earns its place), verification of hallucinated claims and hallucinated or malformed bibliography entries, the ACL policy on AI writing assistance and what must be disclosed, and best practices for paper structure, experiments, reproducibility checklists, and rebuttals. Use this skill whenever the user is drafting, editing, polishing, restructuring, or reviewing a paper, abstract, intro, related work, rebuttal, or camera-ready; whenever they ask to "clean up" or "tighten" academic writing; whenever citations or a .bib file need checking; and whenever they ask what has to be disclosed about AI assistance. Reach for it even when the user only says "make this section better", "check my references", or "is this ready to submit".
The English summary is being prepared. Showing the original description.
Generates publication-quality figures for ML papers from research context. Given a paper section or description, extracts system components and relationships to generate architecture diagrams via Gemini. Given experiment results or data, auto-selects chart type and generates data-driven figures via matplotlib/seaborn. Use when creating any figure for a conference paper.
The English summary is being prepared. Showing the original description.
Use when deciding whether a project is a strong AISTATS fit, comparing AISTATS with NeurIPS, ICML, ICLR, UAI, COLT, JMLR, statistics journals, or application venues, identifying the statistical primitive of the contribution, and sharpening the AI-statistics framing before writing begins.
The English summary is being prepared. Showing the original description.
Use when deciding whether a project is a genuine ACM MM (ACM Multimedia) contribution rather than single-modality work, choosing a thematic area, and routing between ACM MM, CVPR/ICCV, ACL/EMNLP, ICMR, MMSys, NeurIPS/ICLR, and the ACM TOMM journal by finding the cross-modal or media-systems core of the contribution.
The English summary is being prepared. Showing the original description.
Use when deciding whether a project is a strong AAAI submission across its broad AI scope, should be reframed or routed to a dedicated track such as AI for Social Impact or AI Alignment, or should instead go to IJCAI, NeurIPS, ICML, ICLR, AISTATS, UAI, ACL, CVPR, KDD, CHI, ICRA, or another specialist venue.
The English summary is being prepared. Showing the original description.
Use when positioning an AAAI paper's novelty against archival work, contemporaneous arXiv or workshop papers, and AAAI/IJCAI/NeurIPS/ICML/ICLR neighbors across the broad AI scope, while staying inside AAAI's dual-submission and AI-as-source policy constraints and writing a related-work section legible to non-specialist reviewers.
The English summary is being prepared. Showing the original description.
Use when positioning an AAMAS submission against multiagent, game-theory, and reinforcement-learning literature spread across AAMAS, AAAI, IJCAI, NeurIPS, ICML, EC, and JAAMAS, including arXiv and workshop versions, concurrent submissions, prior conference versions, and the cross-community citation coverage AAMAS reviewers expect.
The English summary is being prepared. Showing the original description.
Use when deciding whether a project is a strong AAMAS fit, comparing AAMAS with AAAI, IJCAI, NeurIPS, ICML, EC, and the JAAMAS journal, identifying whether the agents are truly the research object, naming the interaction primitive (solution concept, mechanism, coordination, negotiation), and sharpening the multiagent framing before writing begins.
The English summary is being prepared. Showing the original description.
机器之心/Synced（国内首家系统性AI科技媒体）的AI内容创作思维——论文级深度+量化驱动+学术与产业双线并跑+产品化媒体运营。 触发词：「机器之心视角」「像机器之心那样写」「Synced Review」「AI论文解读」「SOTA评测」「AI产业分析」「GMIS大会」。 擅长：顶会论文拆解报道（ICLR/NeurIPS/CVPR）、AI模型数据评测（SOTA平台）、产业趋势三层推演（技术→商业→生态）、AI中国年度评选、学术中立+产业判断、PRO会员通讯。
The English summary is being prepared. Showing the original description.
End-to-end academic paper writing workflow. Takes a research repository and produces a publication-ready LaTeX manuscript for top ML/AI conferences (NeurIPS, ICML, ICLR, ACL, AAAI, COLM). Covers repo understanding, structure planning, section drafting, citation management, anti-AI polishing, and conference formatting. Triggers on "write paper", "draft paper", "scholar writing", "paper writing workflow".
The English summary is being prepared. Showing the original description.
Help authors learn from exemplary ICLR, ICML, and NeurIPS papers through source-linked manuscript comparisons, concrete writing and experiment suggestions, and guided reflection. Use for research storytelling, contribution framing, paper comparison, figure planning, or research retrospectives. Not an acceptance predictor or a routine grammar/citation formatter.
The English summary is being prepared. Showing the original description.
Think and work like an expert Machine Learning Researcher. Use when a task calls for Machine Learning Researcher judgment. Reasons from population risk, double descent, and inductive bias; enforces sacred test sets, hierarchical ablations, nested CV, and HELM/Dynabench-aware benchmarking; reports with NeurIPS and Pineau reproducibility checklists while treating leakage, meta-overfitting, benchmark contamination, Goodhart gaming, and seed variance as first-class failure modes.
The English summary is being prepared. Showing the original description.
Think and work like an expert Deep Learning Scientist. Use when a task calls for Deep Learning Scientist judgment. Reasons from CNN/Transformer inductive bias, Li et al. loss landscapes, grokking/mode connectivity, and Kaplan/Chinchilla scaling (~20 tokens/param); designs ResNet/ViT/DiT/MoE/FlashAttention stacks with FLOPs-matched ablations; trains AdamW+cosine/WSD via Megatron-FSDP/DeepSpeed; evaluates FID/MMLU-Pro/MMLU-CF with lm-eval decontamination and Pineau/NeurIPS reproducibility checklists.
The English summary is being prepared. Showing the original description.
Think and work like an expert AI Researcher. Use when a task calls for AI Researcher judgment. Reasons from data generating processes, inductive biases, and compute-data-algorithm trade-offs through train/val/test discipline, seed sweeps, ablation ladders, and standards like NeurIPS reproducibility checklists, model cards, and lm-eval-harness, while treating data leakage (Kapoor & Narayanan taxonomy), benchmark contamination, spurious correlations, and reward-model overoptimization as first-class failure modes.
The English summary is being prepared. Showing the original description.
Base template for Stage1 reviewer-comment breakdown. Use when creating or extending conference-specific Stage1 skills (for example ICLR, ICML, NeurIPS, ACL) so shared extraction, splitting, and response-mapping logic stays consistent.
The English summary is being prepared. Showing the original description.
Generate, critique, and iterate image-generation prompts for high-fidelity academic paper poster reference mockups before implementing HTML/PDF/PPTX. Use when Codex is asked to create a poster concept image, imagegen prompt, visual reference draft, layout mockup, or "生图" design for ACL/EMNLP/NeurIPS-style A0 posters from a paper, figures, metrics, logos, QR codes, or an existing poster screenshot.
The English summary is being prepared. Showing the original description.
Simulate rigorous, fair peer review on a paper — in two modes. (A) Red-team YOUR OWN draft before submission to predict the reviews you'll get and get a fix list. (B) Review SOMEONE ELSE'S paper when you're an assigned reviewer or helping your advisor — produce a fair, venue-formatted, submission-ready review. Use whenever the user gives a paper (PDF, .tex, .md, arXiv link, or pasted text) and wants to predict reviews, find weaknesses before reviewers do, stress-test a submission, OR write a real review of another paper. Grounded in official NeurIPS/ICLR/ACL reviewer guidelines; every criticism pinned to a location, and checked against ACL's H1–H17 list of illegitimate critiques. 两种模式:审自己的稿(投稿前红队)/ 审别人的稿(当审稿人或帮导师),按会议官方标准评,绝不脑补、绝不发不正当批评。
The English summary is being prepared. Showing the original description.
Use when independently reviewing a research-paper draft before submission, running a mock top-conference panel, assessing readiness or accept/reject risk, or predicting reviewer concerns for venues such as CVPR, ICCV, ECCV, ICLR, NeurIPS, and AAAI. Does not draft or audit author responses after official reviews arrive.
The English summary is being prepared. Showing the original description.