From understanding research to finding papers, experimenting, writing, and communicating. Choose material for what you are doing now, then return to the corresponding workflow page.
Understanding research: reading, thinking, and the unknown
Why are research skills so seldom taught? — Michael Nielsen — An English essay drawing on physics research to compare research skills with classroom problem-solving and discuss how researchers develop a way of working.
Advice to a Beginning Graduate Student — Manuel Blum — English advice on reading, thinking, research, and writing. Begin with top-down reading: skip ahead around a question, then fill knowledge gaps.
The importance of stupidity in scientific research — Martin A. Schwartz — A one-page English essay from 2008 about moving from questions with known answers into genuine uncertainty. Useful when beginning research.
How Do I Get Started in Research? — Stanford Academic Advising — Undergraduate advice on starting from interests, faculty conversations, and projects. Institutional programs are intended for Stanford students.
Technical questions and issue reports
- How To Ask Questions The Smart Way, Chinese translation — Technical communication advice. Consult how to explain goals, attempts, environment, and evidence when writing a help request or GitHub issue.
Finding papers, tracing sources, and saving material
Google Scholar Search Help — Official help for searching by author and title, following citations and related papers, and using search syntax.
AMiner platform guide — Chinese product documentation for scholarly search, Q&A, and reading. Learn the features and follow citations back to papers.
Zotero: The Basics — Official introduction to saving, organizing, and citing sources. Start with collections, items, and attachments.
OpenReview documentation — Official documentation for paper versions, public reviews, author responses, and related functions.
Hugging Face Daily Papers — An entry to daily papers and discussion. Follow interesting work to the original paper, code, and project.
Obsidian｜Local Markdown notes and links between papers, questions, and projects.
Zotero Integration｜Import Zotero citations, notes, and PDF annotations into Obsidian; setup is documented in the repository.
Zotero Connector｜Save bibliographic information and available attachments from your browser.
Scholaread — Paper reading, translation, and notes.
切问 / Qiewen｜Search for papers from a research question, then read and compare related work.
Reading a first paper and understanding publication
How to Read a Paper — S. Keshav — A two-page English guide to three-pass reading: establish what a paper does, then decide how deeply to read. Try it with a paper beside you.
Mu Li: paper-reading index — Chinese close-reading videos. Start with how to read a paper, then watch how he examines an abstract, figures, experiments, and structure in a paper you are studying. Transformer close reading
ACL Anthology — A natural-language-processing paper archive. Search by title, author, or venue for papers, versions, and citation information.
Proceedings of Machine Learning Research — ML conference and workshop proceedings. Find formal papers, supplementary material, and related work from the same event.
Experiments, tuning, diagnosis, and evaluation
A Recipe for Training Neural Networks — Karpathy — English training experience from 2019. A diagnosis sequence built around inspecting data, starting simply, and checking fit.
Deep Learning Tuning Playbook — Google Research — An English tuning handbook. Consult initial setup, experiment organization, and optimization once you are running experiments.
scikit-learn: Common pitfalls and recommended practices — Official examples of leakage, inconsistent preprocessing, and randomness, with incorrect and recommended approaches side by side.
Task-Specific LLM Evals that Do & Don't Work — Eugene Yan — An English practical blog on choosing what and how to evaluate for a specific LLM task. Useful when an application already exists and needs evaluation.
Jupyter｜Keep code, analysis, tables, and figures together in a notebook.
Rainbow CSV｜Distinguish CSV/TSV columns and check column consistency in VS Code.
Writing, LaTeX, Markdown, and presentation
How to write a great research paper — Simon Peyton Jones — English talk and slides on communicating the problem, contribution, and argument. Useful while structuring a paper.
MLNLP Paper-Writing-Tips — A Chinese writing checklist for checking format, expression, and submission details once you have a draft.
Learn LaTeX in 30 minutes — Overleaf — An introductory tutorial covering text, equations, figures, and citations before applying a venue template.
Markdown Guide: Basic Syntax — A syntax reference for headings, links, lists, images, and code blocks, useful for documents and GitHub READMEs.
How to give a great research talk — Simon Peyton Jones and others — Talks and slides, including a Chinese slide entry, on organizing information around the audience and planning a research presentation.
Fundamentals of Data Visualization — Claus O. Wilke — An open illustrated book. Consult chart selection, colors, and readability for the figure you are making.
AstraDraw — A collaborative whiteboard and presentation project marked Alpha in its repository. Try it for organizing diagrams and presentations.
LaTeX Workshop｜Build and preview LaTeX and complete citations in VS Code with a local TeX installation.
Code Spell Checker｜Check English spelling in notes, READMEs, and code comments.
AI in research: tasks, checks, and reuse
Building effective agents — Anthropic — An English engineering article from 2024 comparing fixed workflows and autonomous agents, and extending simple structures gradually.
How I use LLMs — Karpathy — An English video demonstrating LLM use in everyday tasks; a concrete collection of tool-use examples.
Building A Generative AI Platform — Chip Huyen — An English engineering blog connecting data, models, retrieval, and evaluation from a systems perspective. Read after building a simple generative-AI application.
Full Stack Deep Learning — Courses, talks, and engineering resources covering data, development, and deployment beyond the model. Choose by your current project stage.
Paper-reading toolkit — Paper-reading tools, including Grill Me for questions about methods and evidence.