The annotated PyTorch training loopidlemachines.co.uk81 points·by smaddrellmander·20 dni temu·19 comments
Mae vs. MSE: more than just the mean vs. median debateidlemachines.co.uk1 points·by smaddrellmander·24 dni temu·0 comments
DiffusionGemma: Discrete diffusion in a large language modelidlemachines.co.uk3 points·by smaddrellmander·w zeszłym miesiącu·1 comments
Heaven knows I'm perplexed nowidlemachines.co.uk2 points·by smaddrellmander·w zeszłym miesiącu·0 comments
Reading MAI's efficiency gain. How to pick architectures like serious peopleidlemachines.co.uk9 points·by smaddrellmander·w zeszłym miesiącu·0 comments
MAI-Thinking-1: Building a Hill-Climbing Machine [pdf]microsoft.ai3 points·by smaddrellmander·w zeszłym miesiącu·0 comments
Are contrastive losses just cross entropy all along?idlemachines.co.uk2 points·by smaddrellmander·w zeszłym miesiącu·0 comments
Every token, everywhere, all at onceidlemachines.co.uk2 points·by smaddrellmander·2 miesiące temu·0 comments
The cut in the Mixture of Experts compute graphidlemachines.co.uk1 points·by smaddrellmander·2 miesiące temu·0 comments
Softmax, can you derive the Jacobian? And should you care?idlemachines.co.uk134 points·by smaddrellmander·3 miesiące temu·47 comments
Gemma 4 is not your standard transformeridlemachines.co.uk2 points·by smaddrellmander·3 miesiące temu·0 comments
smaddrellmander·4 dni temu·discussLove this! Cute graphics and the way the balls combine is really fun. Great job.
smaddrellmander·2 miesiące temu·discussA 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.