🔄 In a Training Loop

John Zakkam

johnzakkam

3 4

·

https://jzk.vercel.app

AI & ML interests

squeezing neural nets into NPUs so they fit on a diet.

Recent Activity

liked a model 5 days ago

lazybrick/Qwen3.5-4B-Kiln-Glaze-W4A16

updated a model 15 days ago

lazybrick/Qwen3.5-4B-Kiln-AutoRound-W4A16-g128

updated a dataset 15 days ago

lazybrick/kiln-evals

View all activity

Organizations

upvoted a collection 16 days ago

Kiln: Qwen3.5-4B, fired small

Qwen3.5-4B compressed with GPTQ, AWQ and INT8, each scored against BF16 under one fixed protocol. • 7 items • Updated 15 days ago • 2

upvoted a paper about 2 years ago

Can we learn better with hard samples?

Paper • 2304.03486 • Published Apr 7, 2023 • 1

upvoted a collection about 2 years ago

Llama 3.2

This collection hosts the transformers and original repos of the Llama 3.2 and Llama Guard 3 • 15 items • Updated Dec 6, 2024 • 678