Forem

Lewis Won profile picture

Lewis Won

A software engineer with interest in LLMs

Location Singapore Joined Joined on  github website
How does low-rank adaptation for large language models work
Cover image for How does low-rank adaptation for large language models work

How does low-rank adaptation for large language models work

4
Comments
21 min read

Want to connect with Lewis Won?

Create an account to connect with Lewis Won. You can also sign in below to proceed if you already have an account.

Already have an account? Sign in
FlashAttention by hand
Cover image for FlashAttention by hand

FlashAttention by hand

4
Comments
23 min read
Online softmax by hand
Cover image for Online softmax by hand

Online softmax by hand

Comments
20 min read
Tensor parallelism by hand
Cover image for Tensor parallelism by hand

Tensor parallelism by hand

7
Comments
28 min read
Routing and balancing losses with Mixture of Experts
Cover image for Routing and balancing losses with Mixture of Experts

Routing and balancing losses with Mixture of Experts

4
Comments
17 min read
Plain Guide to Einops
Cover image for Plain Guide to Einops

Plain Guide to Einops

1
Comments
16 min read
ZeRO by hand with a 4-parameter model
Cover image for ZeRO by hand with a 4-parameter model

ZeRO by hand with a 4-parameter model

1
Comments 1
23 min read
From Scatter to All-Reduce: A Plain-English Guide to Collective Operations
Cover image for From Scatter to All-Reduce: A Plain-English Guide to Collective Operations

From Scatter to All-Reduce: A Plain-English Guide to Collective Operations

7
Comments
21 min read
Demystifying GPUs: From Core Architecture to Scalable Systems
Cover image for Demystifying GPUs: From Core Architecture to Scalable Systems

Demystifying GPUs: From Core Architecture to Scalable Systems

81
Comments 2
12 min read
Key Insights Gained from Building and Training LLM
Cover image for Key Insights Gained from Building and Training LLM

Key Insights Gained from Building and Training LLM

3
Comments
27 min read
Contextual chunking for Retrieval Augmented Generation
Cover image for Contextual chunking for Retrieval Augmented Generation

Contextual chunking for Retrieval Augmented Generation

2
Comments
59 min read
Implementing DeepSeek-R1 GRPO in Apple MLX framework
Cover image for Implementing DeepSeek-R1 GRPO in Apple MLX framework

Implementing DeepSeek-R1 GRPO in Apple MLX framework

12
Comments
29 min read
Efficient self-attention mechanism
Cover image for Efficient self-attention mechanism

Efficient self-attention mechanism

1
Comments 1
31 min read
Creating the self-attention mechanism from scratch
Cover image for Creating the self-attention mechanism from scratch

Creating the self-attention mechanism from scratch

3
Comments
22 min read
Navigating the world of Harry Potter with Knowledge Graphs
Cover image for Navigating the world of Harry Potter with Knowledge Graphs

Navigating the world of Harry Potter with Knowledge Graphs

Comments 2
6 min read
loading...