Browse and solve 1,200+ machine learning coding challenges on Deep-ML. Filter by difficulty (beginner, intermediate, advanced) and category (linear algebra, machine learning, deep learning, NLP, computer vision). Write Python solutions in the browser and get instant feedback against test cases.
Deep-ML | Conditional Flow Matching with Gaussian PathsConditional Flow Matching with Gaussian Paths
At t = 0.5 with sigma_min = 0.1 the path has sigma_t = 1 - 0.9 * 0.5 = 0.55, so x_t = 0.55 * x0 + 0.5 * x1: the first row is [0.5, 1.0] and the second is [1.1 + 0.5, 1.0] = [1.6, 1.0]. The target velocity is x1 - 0.9 * x0, giving [1, 2] and [-0.8, 2]; with a zero prediction the per-sample losses are the squared norms 5 and 4.64.
Lipman et al. (2023) train a velocity field by regressing onto the velocity of a simple conditional probability path pt(x∣x1)=N(μt(x1),σt(x1)2I). Implement the general velocity formula, the optimal-transport path, the variance-preserving diffusion path and the loss.