MagicaLabs details its pretraining recipe: tens of multiplicative changes, bug fixes as compute multipliers
magicailabs · x · 2026-09-09
MagicaLabs explains its pretraining recipe is the multiplicative result of tens of changes across model architecture, optimizer, training objective, and data — and that fixing minor bugs proved to be a compute multiplier too. More on the research process in their blog.
Related event: Magic Claims 50x Pretraining Efficiency, Matching DeepSeek V4 Pro at ~$500K(7 posts)→
More from Research
- Proof drama escalates: releasing details would expose the method, argues critic — Afinetheorem · 2026-09-09
- 105 hours, 4.9M messages, 300B output tokens: AI proves previously unsolved math problem — zainhas · 2026-09-09
- Mathematicians blast Navier-Stokes blowup paper for omitting Chen-Hou, Elgindi and other key references — snikolov · 2026-09-09
- OpenAI claims huge Navier-Stokes math discovery, academics cry foul — wiredmagazine · 2026-09-09
- Strogatz: a Navier-Stokes singularity wouldn't invalidate the model, just end the continuum approximation — stevenstrogatz · 2026-09-09
- Vortex stretching vs viscosity: the battle inside Navier-Stokes that could breed a singularity — stevenstrogatz · 2026-09-09