Acing AI — AI education, tutorials, research and datasets for data scientists
The Autonomous AI Agent Cyberattack Just Said It Was Authorized: Refusal Is Not a Security Boundary
Two autonomous AI agent cyberattack cases bypassed safeguards by claiming an authorized pentest. Why refusal is a distribution, and where authorization belongs.
Latest Intelligence
Curated technical papers and hands-on implementation guides for the modern AI engineer.
Agent Skills Are a Procedure, Not Knowledge
A controlled study of agent skills finds they work by procedural anchoring, not knowledge injection. What that means for your SKILL.md and your skill library.
LLM Capacity Planning Has a Half-Life: Why Your Serving Numbers Expire
LLM capacity planning has a half-life: the VRAM bill and throughput baseline you wrote at model release decay as vLLM and SGLang change defaults beneath them.
ArticleEvolution Strategies Fine-Tuning: What Dropping Backprop Buys, and Where the Bill Moves
ArticleVision Encoder Disaggregation: When Your Encoder Should Be Its Own Service
TutorialHow Much Does an Image Cost? Measuring Image Token Cost on a 27B Multimodal Model
Browse by Type
Tutorials
Step-by-step guides from neural network basics to advanced LLM fine-tuning.
Research Papers
Peer-reviewed insights and white papers defining the frontier of artificial intelligence.
Datasets
High-fidelity training sets for natural language processing and computer vision.
Start Learning
Guided sequences through our best content — structured to build understanding from the ground up.
Post-Training Modern LLMs
Pretraining produces a model that predicts text. Post-training is what turns it into something you can ship. This path walks the levers in the order you would actually reach for them: supervised fine-tuning and adapters, preference optimization without a reward model, the reinforcement-learning map from RLHF to verifiable rewards, RL against a verifier that cannot be talked out of its answer, and finally the inference-time compute that picks up where training leaves off. Every step names the ceiling it runs into.
Evaluating LLMs Honestly
A leaderboard number is a hypothesis, not a result. This path builds the habit of asking what a benchmark measured before quoting what it reported, starting with contamination and judge bias, moving through a case where the advertised figure and the measured one diverge, then to agents where a single passing run tells you almost nothing, and ending where the eval harness itself turns into attack surface.