Short stories 1 - Model merging

Machine-learning models are often fine-tuned for specific tasks from a pre-trained base model. One model might perform well on a language task, another on an image problem, and another on a different domain altogether. Here's Bethan on model merging.

79.28 A Casio fx-7700GB cubics program
 Ardakov, K The Mathematical Gazette volume 79 issue 485 352-353 (01 Jul 1995)
81.44 Powers and other functions of 2 × 2 matrices
 Ardakov, K The Mathematical Gazette volume 81 issue 492 434-437 (01 Nov 1997)
Randomized Milstein algorithm for approximation of solutions of jump–diffusion SDEs
 Przybyłowicz, P Schwarz, V Szölgyenyi, M Journal of Computational and Applied Mathematics volume 440 115631-115631 (Apr 2024)
A Skorohod measurable universal functional representation of solutions to semimartingale SDEs
 Przybyłowicz, P Schwarz, V Steinicke, A Szölgyenyi, M Stochastic Analysis and Applications volume 42 issue 6 1137-1155 (08 Nov 2024)
A higher-order approximation method for jump-diffusion SDEs with a discontinuous drift coefficient
 Przybyłowicz, P Schwarz, V Szölgyenyi, M Journal of Mathematical Analysis and Applications volume 538 issue 1 128319-128319 (Oct 2024)
Lower error bounds and optimality of approximation for jump-diffusion SDEs with discontinuous drift.
 Przybyłowicz, P Schwarz, V Szölgyenyi, M BIT Numer Math volume 64 issue 4 35 (2024) https://www.ncbi.nlm.nih.gov/pubmed/39263602
IIB an equivariantly localized puncture
 Sparks, J Couzens, C Luscher, A SciPost Physics
Strong convergence to operator-valued semicirculars
 Pi, J Probability theory and related fields
Subscribe to