Wei Yi – Medium

Pinned

Wei Yi
in
Towards Data Science

How does the Segment-Anything Model’s (SAM’s) decoder work?

A deep dive into how the Segment-Anything model’s decoding procedure, with a focus on how its self-attention and cross-attention mechanism…

18 min readMar 24, 2024

--

1

How does the Segment-Anything Model’s (SAM’s) decoder work?

--

1

Pinned

Wei Yi
in
Towards Data Science

Speeding up vision transformer prediction by 9 times faster with PyTorch, ONNX and TensorRT

How to use 16bit float, TensorRT, network rewriting and multi-threading to dramatically speed up deep learning model prediction

11 min readJun 4, 2023

--

Speeding up vision transformer prediction by 9 times faster with PyTorch, ONNX and TensorRT

--

Wei Yi
in
Towards Data Science

How Decision Trees Split Nodes, from Loss Function Perspective

Learn how a decision tree splits nodes only to minimize its loss function

12 min readMay 15, 2023

--

1

How Decision Trees Split Nodes, from Loss Function Perspective

--

1

Wei Yi
in
Towards Data Science

Distributed data parallel and distributed model parallel in PyTorch

How distributed data parallel DDP and distributed model parallel DMP works in stochastic gradient descent with large models and huge data

14 min readMay 8, 2023

--

1

Distributed data parallel and distributed model parallel in PyTorch

--

1

Wei Yi
in
Towards Data Science

Understanding the Denoising Diffusion Probabilistic Model, the Socratic Way

A deep dive into the motivation behind the denoising diffusion model and detailed derivations for the loss function

69 min readFeb 25, 2023

--

5

Understanding the Denoising Diffusion Probabilistic Model, the Socratic Way

--

5

Wei Yi
in
Towards Data Science

The Input-output Attention Mechanism from “Neural Machine Translation by Jointly Learning…

Learn the math and intuition behind the input-output attention mechanism in a RNN-based language to language translation model

11 min readMar 18, 2022

--

The Input-output Attention Mechanism from “Neural Machine Translation by Jointly Learning…

--

Wei Yi
in
Towards Data Science

Can We Use Stochastic Gradient Descent (SGD) on a Linear Regression Model?

Learn why it is valid to use SGD on a linear regression model for parameter learning, see however, SGD can be inefficient, and appreciate…

17 min readAug 5, 2021

--

Can We Use Stochastic Gradient Descent (SGD) on a Linear Regression Model?

--

Wei Yi
in
Towards Data Science

Where do confidence interval in linear regression come from — the case of least squares formulation

This article explains in least square linear regression model, how to understand parameter std err, t, P>|t| and confidence intervals.

21 min readJun 28, 2021

--

Where do confidence interval in linear regression come from — the case of least squares formulation

--

Wei Yi
in
Towards Data Science

Sparse and Variational Gaussian Process — What To Do When Data is Large

Learn how the Sparse and Variational Gaussian Process model uses inducing variables to scale to large datasets.

47 min readJun 26, 2020

--

10

Sparse and Variational Gaussian Process — What To Do When Data is Large

--

10

Wei Yi
in
Towards Data Science

Variational Gaussian Process — What To Do When Things Are Not Gaussian

Learn to use non-Gaussian distributions in Gaussian Process models, and variational inference with Gaussian quadrature to compute…

55 min readJun 5, 2020

--

5

Variational Gaussian Process — What To Do When Things Are Not Gaussian

--

5

Wei Yi

Wei Yi

I am a principal data scientist at AstraZeneca. Previously I worked at SecondMind, Microsoft Research, and also was CTO of a hedge fund EQB.

Following

Help
Status
About
Careers
Blog
Privacy
Terms
Text to speech
Teams