MD Sunbeam | WritingTechnical essays about evaluation, learning systems, and the assumptions that determine their behavior.https://mdsunbeam.comen-usWhat an Evaluation Harness Decideshttps://mdsunbeam.com/blog/evaluation-harnesshttps://mdsunbeam.com/blog/evaluation-harnessThe model is only one part of a closed-loop benchmark. Parsing, recovery, timing, and rollout accounting determine what its score means.Sat, 15 Aug 2026 00:00:00 GMTWhat a Convolution Assumeshttps://mdsunbeam.com/blog/unraveling_convolutionshttps://mdsunbeam.com/blog/unraveling_convolutionsConvolution is a wager: nearby values interact, and one rule should work everywhere. That explains both its efficiency and its blind spots.Mon, 13 May 2024 00:00:00 GMTOne ReLU Neuron, Four Learning Rateshttps://mdsunbeam.com/blog/simple_beautiful_nnshttps://mdsunbeam.com/blog/simple_beautiful_nnsA one-line recurrence separates smooth convergence, oscillation, a two-cycle, and a dead ReLU.Sat, 13 Apr 2024 00:00:00 GMT