Retrieval - Bring the Broken Part
Word matching and embeddings fail on opposite queries, so production search runs both lookups and merges them using nothing but the ranks.
Gaussian Splatting - Three Million Balloons
A room can be stored as three million fuzzy blobs instead of a mesh or a network, and the whole renderer is a depth sort and some tinted glass.
Voice Models - Press to Talk
A model that can hear and speak still cannot hold a conversation. The hard part is knowing when to start, and being willing to stop.
Voice Models - Writing Down a Concert
Sound has to become a short list of symbols before a language model can speak it. The method that works writes each frame down in eight passes.
Test-Time Compute - Order One More Test
A model can be made better after it is trained, just by letting it answer for longer. Almost all of that gain comes from the thing that checks the answer.
Tokenizers - Out of Sorts
A model's vocabulary is fitted by counting pairs and frozen before training starts, and it decides what the model can spell, add up and charge you.
LLM Benchmarks - The Horse and the Rider
The famous benchmarks saturated and leaked. The agentic ones replacing them score a model and the code driving it together, and cannot tell the two apart.
MCTS and AlphaZero - Four Steps in the Dark, Part 2
AlphaGo gave tree search a policy and a value network. AlphaGo Zero made the search the network's teacher, in an update that sits close to PPO's.
MCTS and AlphaZero - Four Steps in the Dark, Part 1
Why game programs search a tree, why they fill it with random games, and what plain Monte Carlo tree search does on real Go and chess positions.
Jev - The Classifier Strikes Back
Jev returns typed, calibrated decisions instead of text. Underneath, it looks like a very good zero-shot classifier, and calibration is the claim to test.
Mamba - Is Attention All You Really Need? Part 2
Selection broke the convolution that made state space models fast to train. Here is how Mamba got the speed back, and why attention stayed in the room.
Mamba - Is Attention All You Really Need? Part 1
State space models come from control theory, run like an RNN and train like a CNN. Mamba's one change lets each word decide what the model remembers.