HNHacker News
TopNewBestAskShowJobs

smaddrellmander

97 karma · joined April 21, 2026

idlemachines.co.uk
submissionscomments

Fast weights and sparse attention in GLM-5.3-Flash

idlemachines.co.uk·7 pts·smaddrellmander·
0

Tokenisation: Who decides what a token is anyway?

idlemachines.co.uk·2 pts·smaddrellmander·
0

Adam and AdamW: adaptive optimisation and weight decay

idlemachines.co.uk·1 pts·smaddrellmander·
0

Thinking Machines dropped RoPE, and it's a good idea

idlemachines.co.uk·6 pts·smaddrellmander·
0

Nesso-1: Accelerating Open-Source Binding Affinity Predictions [pdf]

valencelabs.com·2 pts·smaddrellmander·
1

The annotated PyTorch training loop

idlemachines.co.uk·81 pts·smaddrellmander·
19

Mae vs. MSE: more than just the mean vs. median debate

idlemachines.co.uk·1 pts·smaddrellmander·
0

DiffusionGemma: Discrete diffusion in a large language model

idlemachines.co.uk·3 pts·smaddrellmander·
1

Heaven knows I'm perplexed now

idlemachines.co.uk·2 pts·smaddrellmander·
0

Reading MAI's efficiency gain. How to pick architectures like serious people

idlemachines.co.uk·9 pts·smaddrellmander·
0

MAI-Thinking-1: Building a Hill-Climbing Machine [pdf]

microsoft.ai·3 pts·smaddrellmander·
0

Are contrastive losses just cross entropy all along?

idlemachines.co.uk·2 pts·smaddrellmander·
0

Every token, everywhere, all at once

idlemachines.co.uk·2 pts·smaddrellmander·
0

The cut in the Mixture of Experts compute graph

idlemachines.co.uk·1 pts·smaddrellmander·
0

DeepSeek V4 from the Inside

idlemachines.co.uk·2 pts·smaddrellmander·
0

Softmax, can you derive the Jacobian? And should you care?

idlemachines.co.uk·134 pts·smaddrellmander·
47

Gemma 4 is not your standard transformer

idlemachines.co.uk·2 pts·smaddrellmander·
0