Kyurae Kim

icon indicating a website link krkim.me icon indicating an email [email protected]

icon company University of Pennsylvania icon location Philadelphia, United States icon location PhD student at the University of Pennsylvania researching Bayesian inference methods. Also interested in optimization, signal processing, HPC, Julia

Results 145 comments of Kyurae Kim

AMAGOLD support

Stochastic gradient MCMC methods, in general, are very young, and radically new approaches emerge every year. Unless AMAGOLD specifically proves to be real-world ready, I doubt that it would earn...

Zygote error differentiating `Coupling`

@devmotion Unfortunately, I can't wind up to older versions of `Zygote` due to an issue with Flux needing `OneElement` which seems to not exist anymore. Any idea what might have...

Zygote error differentiating `Coupling`

I guess I'll have to try `ReverseDiff` for the meantime.

ScikitLearn declared inside a module causes segmentation error

For anyone who's still wondering here's a snippet that worked for me. Note that the loading is done by PyCall but the rest is done using the `ScikitLearn.jl` API. ```julia...

Make an efficient CUDA microbenchmark framework

working on this on a separate repository https://github.com/MGfoundation/mgbench

Implement Fast Fourier Transform CUDA kernel

@mujjingun do you consider this issue solved?

Implement Fast Fourier Transform CUDA kernel

@mujjingun ok

Implement Fast Fourier Transform CUDA kernel

@mujjingun cuFFT have some serious 'warming up' issues. On multi-GPU servers constructing even takes 1000ms or so. I think implementing our own parallel FFT with upsampling built in, is a...

Different order of parameters in MCMC chain vs. MAP result

I'll close this since this seems to have been resolved.

Kernel dies with Gibbs or ADVI with models involving categorical or dirichlet

I'll close this due to inactivity.