Top New Best Ask Show Jobs

gpjt

1,810 karma · joined January 12, 2009

https://www.gilesthomas.com/

submissions comments

Extending Raschka's GPT-2: an MoE trained from scratch on an RTX 3090

gilesthomas.com·1 pts·gpjt·

Putting my Jax-trained models on the Hugging Face Hub

gilesthomas.com·2 pts·gpjt·

Why do OpenAI's GPT-2 weights beat mine? Part four: digging into dropout

gilesthomas.com·1 pts·gpjt·

First patient to undergo live AI-assisted brain surgery has tumour removed

bbc.com·6 pts·gpjt·

Adding diagrams to my static site generator with D2

gilesthomas.com·3 pts·gpjt·

Use the built-in GELU, don't roll your own

gilesthomas.com·1 pts·gpjt·

A Quick(ish) Chinchilla Check

gilesthomas.com·1 pts·gpjt·

I use AI on this blog

gilesthomas.com·1 pts·gpjt·

Why do OpenAI's GPT-2 weights beat mine? Part three: testing overtraining

gilesthomas.com·2 pts·gpjt·

Why do OpenAI's GPT-2 weights beat mine? Part two: the bugfix

gilesthomas.com·8 pts·gpjt·

Why do OpenAI's GPT-2 weights beat mine?

gilesthomas.com·4 pts·gpjt·

Benchmarking Qwen 3.6 35B MoE (3B active) on an RTX 3090

gilesthomas.com·16 pts·gpjt·

Building intuition about LLM parameter counts

gilesthomas.com·2 pts·gpjt·

Poppy the training box, part 1: the beginnings

gilesthomas.com·3 pts·gpjt·

From bigrams to GPT-2, one component at a time (in Jax)

gilesthomas.com·1 pts·gpjt·

Building a Jax training loop for an LLM training run

gilesthomas.com·2 pts·gpjt·

Thoughts on Role Confusion

gilesthomas.com·3 pts·gpjt·

Flax debugging: making a hash of things

gilesthomas.com·2 pts·gpjt·

10Gb/s Ethernet: switching to a Broadcom SFP+ module

gilesthomas.com·195 pts·gpjt·

Jax: Commitment Issues

gilesthomas.com·4 pts·gpjt·

Jax Back Ends and Devices

gilesthomas.com·2 pts·gpjt·

Using Safetensors with Flax

gilesthomas.com·2 pts·gpjt·

First Looking into Jax

gilesthomas.com·3 pts·gpjt·

10Gb/s Ethernet: using mini-heatsinks with a 10GBASE-T SFP+ module

gilesthomas.com·3 pts·gpjt·

10Gb/s Ethernet: what I did to get it working in my home

gilesthomas.com·232 pts·gpjt·

10Gb Ethernet: what I had to (re)learn

gilesthomas.com·1 pts·gpjt·

LLM from scratch, part 33 – what I learned from the appendices

gilesthomas.com·5 pts·gpjt·

LLM from scratch (32l) – Interventions: updated instruction fine-tuning results

gilesthomas.com·1 pts·gpjt·

How an LLM becomes more coherent as we train it

gilesthomas.com·3 pts·gpjt·

LLM from scratch, part 32k – Interventions: gradient accumulation

gilesthomas.com·2 pts·gpjt·

Page 1 of 4Next →