Hi. I have just started to learn how to use pymc3. Can NUTS of pymc3 work with stochastic gradient? I am thinking about creating a new theano operation that provides a noisy estimation of the gradient and value of the potential energy function. That means when NUTS ask my operation for the gradien…

Probably not… unless the mini-batches are sufficiently large. Might need an extra friction term to damp the noise introduced by mini-batch. [image] Minibatch and NUTS Questions Am I correct in saying that the minibatch mode doesn’t work with NUTS?

Can NUTS work with stochastic gradient / mini-batch?

Questions

Hamster_on_wheels May 2, 2019, 3:20pm 2

Probably not… unless the mini-batches are sufficiently large.

Might need an extra friction term to damp the noise introduced by mini-batch.

Topic		Replies	Views
NUTS - FromFunctionOp has no attribute 'grad' Questions theano	2	998	August 3, 2018
Larger dataset linear regression GPU memory problems with NUTS for larger N and D (batch training?) Questions theano	2	1841	October 3, 2017
Minibatch and NUTS Questions	11	3252	June 22, 2018
What is pm.Potential? Questions	4	2926	May 17, 2019
Auto-diff when using NUTS sampler with blackbox likelihoods Questions development	1	835	September 11, 2020

Can NUTS work with stochastic gradient / mini-batch?

Related topics