smaddrellmander
smaddrellmander on Hacker News
Efficient Streaming Language Models with Attention Sinks
Learning Without Backpopagation
How to quickly get prepared for your technical round
Fast weights and sparse attention in GLM-5.3-Flash
Tokenisation: Who decides what a token is anyway?
Adam and AdamW: adaptive optimisation and weight decay
Thinking Machines dropped RoPE, and it's a good idea
Nesso-1: Accelerating Open-Source Binding Affinity Predictions [pdf]
Re:Nesso-1: Accelerating Open-Source Binding Affinity Predictions [pdf]
on Tuesday July 21, 2026 @09:47AM (#48992305)
Curious what people think? Decent binding affinity model out of Valence Labs.
Re:Show HN: Squishy Billiards – A daily pool game with squishy physics
on Wednesday July 08, 2026 @09:22AM (#48831623)
Love this! Cute graphics and the way the balls combine is really fun. Great job.
The annotated PyTorch training loop
Mae vs. MSE: more than just the mean vs. median debate
DiffusionGemma: Discrete diffusion in a large language model
Heaven knows I'm perplexed now
Reading MAI's efficiency gain. How to pick architectures like serious people
MAI-Thinking-1: Building a Hill-Climbing Machine [pdf]
Are contrastive losses just cross entropy all along?
Every token, everywhere, all at once
Re:Ask HN: What small engineering habits compound the most over time?
on Monday May 25, 2026 @04:14PM (#48271153)
A similar vein, asking why in PRs. Most people don’t do the above, and asking why when it doesn’t make sense really helps understanding what your colleagues are actually thinking. Helps create a unified perspective on the codebase.