crussella0129/LNewq
Text Generation • 5B • Updated • 599
Charles Russella
crussella0129
webbrain-one-555630's profile picture
webbrain-6f3q2's profile picture
JackKozmo29's profile picture·
AI & ML interests
Agentic Coding, Workflow Automation, Harnesses, Loop Engineering, Rapid Prototyping, AI Assisted CAD,
Recent Activity
liked a model 2 days ago
webAI-Official/TwIL-8B reacted to Saccadic-AI's post with 🔥 5 days ago
One neural network. Five mathematical operations. Different learning times.
We trained a small mixture-of-experts transformer on addition, subtraction, multiplication, division, and exponentiation modulo 113.
The operations didn’t generalize together. Multiplication and division reached 90% held-out accuracy first, followed by exponentiation, subtraction, and finally addition.
Multiplication before addition—the reverse of what we expect from a schoolchild.
The accuracy chart in the linked article shows why a single overall accuracy number can hide an interesting learning story.
Measured with IRIS from Saccadic LLC. More figures and analysis: https://huggingface.co/blog/Saccadic-AI/accuracy-plateaued-the-model-kept-changing
Have you observed a similar ordering in multi-task grokking experiments? liked a model 5 days ago
burner22/numbatOrganizations
None yet