statement
EBMs promised a revolution 20 years ago, and the problem was never elegance — it was training at scale.
A statement by מייק. Statements are as reported by the cited publishers — not verified by DecisionGateway.
אבל EBMs הבטיחו מהפכה כבר לפני 20 שנה, והבעיה מעולם לא הייתה האלגנטיות. הבעיה הייתה אימון בסקייל.
Discussion
Related claims
The problem with EBMs is training: it is not enough to teach the model that real examples have low energy; it must also be taught that everything else has high energy, and that 'everything else' is enormous, which is why EBMs were stuck for years with unstable training and expensive sampling.post author · statementDiffusion is almost an EBM: the direction diffusion learns is exactly the gradient-descent direction of some energy (in the transition from noisy data to clean), meaning diffusion directly learns the slope of the energy surface and skips the energy itself — the very trick that bypassed the difficulty of training EBMs.post author · statementIn EBM one can combine several models simply by adding their energies, giving 'both and', whereas in diffusion this works only approximately.post author · statementMIT researchers discovered that despite everyone using MTP, no one noticed a built-in mathematical bug in the training process.post author · investigationThe challenge is to overcome costs, train managers and modernize processes to turn the technology into productivity.post author · statementEmbraer exceeded expectations.post headline · statement
More claims by מייק
In event-driven neuromorphic architectures, parts of the network that do not need to work simply stay at rest.statementIn event-driven neuromorphic architectures, memory is adjacent to the processor.statementIn event-driven neuromorphic architectures, time is a computational feature.statementThe future of AI lies not only in pushing more billions of parameters into hot GPUs, but in a smart combination of classic transformers and LLMs with event-driven neuromorphic architectures.prediction'The end of diffusion' is hype; the better question is whether it is worth paying for more expensive training to get built-in verification and flexible compute at inference time — a real and interesting trade-off.statement
All claims by מייק