Read more at arxiv.org/abs/2508.20883
Including scaling LRW up to image generation with Stable Diffusion 3.5 🐱
Including scaling LRW up to image generation with Stable Diffusion 3.5 🐱
Lattice Random Walk Discretisations of Stochastic Differential Equations
We introduce a lattice random walk discretisation scheme for stochastic differential equations (SDEs) that samples binary or ternary increments at each step, suppressing complex drift and diffusion co...
arxiv.org
August 29, 2025 at 3:07 PM
Read more at arxiv.org/abs/2508.20883
Including scaling LRW up to image generation with Stable Diffusion 3.5 🐱
Including scaling LRW up to image generation with Stable Diffusion 3.5 🐱
As described in the paper, LRW provides multiple benefits but the key motivation for us @normalcomputing.com was the co-design with novel stochastic computing hardware which we believe can drastically accelerate general-purpose SDE sampling.
August 29, 2025 at 3:07 PM
As described in the paper, LRW provides multiple benefits but the key motivation for us @normalcomputing.com was the co-design with novel stochastic computing hardware which we believe can drastically accelerate general-purpose SDE sampling.
Oh you king this is great thanks! I was at Lau Pa Sat the other day but went for shrimp noodles (which were great) because the satay queue was too long
April 28, 2025 at 6:46 AM
Oh you king this is great thanks! I was at Lau Pa Sat the other day but went for shrimp noodles (which were great) because the satay queue was too long
Didn’t listen, good decision
April 27, 2025 at 8:31 AM
Didn’t listen, good decision
However! We’re working on a much broader generalisation of abile which hopefully will be able to share soon 🤞🔜
April 18, 2025 at 4:39 PM
However! We’re working on a much broader generalisation of abile which hopefully will be able to share soon 🤞🔜
Adjacent!
posteriors takes the natural gradient descent viewpoint on EKF arxiv.org/abs/1703.00209
Which is nice for online deep learning but not necessarily bespoke state-space model inference
posteriors takes the natural gradient descent viewpoint on EKF arxiv.org/abs/1703.00209
Which is nice for online deep learning but not necessarily bespoke state-space model inference
Online Natural Gradient as a Kalman Filter
We cast Amari's natural gradient in statistical learning as a specific case of Kalman filtering. Namely, applying an extended Kalman filter to estimate a fixed unknown parameter of a probabilistic mod...
arxiv.org
April 18, 2025 at 4:37 PM
Adjacent!
posteriors takes the natural gradient descent viewpoint on EKF arxiv.org/abs/1703.00209
Which is nice for online deep learning but not necessarily bespoke state-space model inference
posteriors takes the natural gradient descent viewpoint on EKF arxiv.org/abs/1703.00209
Which is nice for online deep learning but not necessarily bespoke state-space model inference
We've also updated the paper and made some cool updates to the library 😎
Paper: arxiv.org/abs/2406.00104
Repo: github.com/normal-compu...
Paper: arxiv.org/abs/2406.00104
Repo: github.com/normal-compu...
April 18, 2025 at 4:01 PM
We've also updated the paper and made some cool updates to the library 😎
Paper: arxiv.org/abs/2406.00104
Repo: github.com/normal-compu...
Paper: arxiv.org/abs/2406.00104
Repo: github.com/normal-compu...
Should have said, here h is stepsize 😅
April 10, 2025 at 9:12 AM
Should have said, here h is stepsize 😅
Yep! That would be clearer
March 25, 2025 at 8:16 PM
Yep! That would be clearer
Thinking about it more, I think the sharp jumps are an artefact of the plotting. The plotting function will linearly interpolate but you can actually probabilistically interpolate using the smoothing equations
February 28, 2025 at 3:15 PM
Thinking about it more, I think the sharp jumps are an artefact of the plotting. The plotting function will linearly interpolate but you can actually probabilistically interpolate using the smoothing equations
Oh you are right! Very nice!
February 28, 2025 at 9:50 AM
Oh you are right! Very nice!
I don’t think so - I can’t see any backward iterations in the code. Also the sharp changes after a result in e.g. the boxing plot are a classic filtering feature - there is a reason smoothing is called smoothing after all 😄
February 28, 2025 at 3:57 AM
I don’t think so - I can’t see any backward iterations in the code. Also the sharp changes after a result in e.g. the boxing plot are a classic filtering feature - there is a reason smoothing is called smoothing after all 😄