jenniferzhao0531/icml2026-guide-cn

ICML 2026 Guide CN is a Chinese-language browsing guide for all 6,567 accepted papers from the ICML 2026 machine learning conference. ICML (the International Conference on Machine Learning) is one of the most influential academic venues for AI research. This project makes it easier to explore that large collection by organizing every paper into a three-level hierarchy and adding a six-dimensional Chinese summary to each one. The tool works in two modes. The first is a ready-made static webpage, a single HTML file with no external dependencies, already published online. It has a three-level collapsible navigation on the left, a full-text search box for titles and authors, and a one-click filter to show only the 575 Spotlight papers (the conference's highest-rated submissions, highlighted with a gold badge). The second mode lets you run the pipeline yourself: four Python scripts crawl the official ICML data, call any OpenAI-compatible language model to classify papers into subcategories and generate Chinese summaries across six dimensions (research motivation, problem being solved, observations, methods, data and experiments, and main contribution), and finally render everything into the HTML file. All LLM calls use an OpenAI-compatible interface, so you can point it at DeepSeek, Claude, or any compatible proxy with your own API key. You would use this if you are a Chinese-speaking researcher or student who wants to scan what was accepted at ICML 2026 without reading 6,567 English abstracts. The project is built with Python for the pipeline and plain HTML for the viewer, with MIT license.