What problem does it solve?
It saves you from manually assembling a high-quality, competition-grade interview prep guide for a specific ML/LLM topic by producing a single long-form Chinese cheat sheet with formulas, derivations, runnable from-scratch PyTorch code, and a full set of 高频面试题.
Core Features & Use Cases
- Topic-scoped, exam-ready tutorials: Converts a chosen ML/LLM topic into a 600–1000 line Chinese tutorial with a fixed section skeleton and L1/L2/L3 question coverage.
- Derivation + from-scratch code together: Includes core formulas with derivations and implements the key ideas in real PyTorch code for review and practice.
- Cross-model review and render pipeline: Performs math/code/factual/citation checks before generating a final academic-style HTML output, suitable for study and sharing.
- Use cases: Prepare for 秋招/面试, create study notes for a specific technology (e.g., KV Cache, DPO/PPO/RLHF, MoE), and generate a durable reference page for later revision.
Quick Start
Use the interview-cheatsheet skill to generate an AI interview tutorial by asking for: 写一份 KV Cache + Speculative Decoding 面试 cheat sheet。