Stories by jxmorris12

Honeycrisp: An Apple-First Deep Learning Framework

My Gear

DOGE for AI

How to Backpack

GRPO with Verifiable Rewards Is Contrastive Loss

Bit Prediction

How Could Telepathy Work?

How to Scale Your Model

RLHF Book

Reading notes: unsupervised word translation

The Art of Debugging

What a $500k grant proposal looks like

Deep Reinforcement Learning Doesn't Work Yet

How far can you get in 40 minutes from each subway station in NYC?

Attention Sinks in LLMs for endless fluency

Flow with What You Know: An Introduction to Flow-Based Models

A History of Nvidia Stream Multiprocessor (2020)

Gaming TruthfulQA: Simple Heuristics Exposed Dataset Weaknesses

Ε, a Nuisance No More

Learning CUDA by Optimizing Softmax

1
2
3
4
5
6
7
8