zeroloss.lol

Hey, I'm Marquis. I build things, train models, and work with computational genomics.

I keep my formal write-ups inside the code repositories, so this site is just a quick, clean directory to learn about my work.

🧬 What I'm Doing Right Now

• Regulatory Genomics: Evaluating sequence-to-function foundation models (like Borzoi and Enformer) using individual-level variation via protected datasets.
• Other Stuff: StatQuest, Deeplizard, 3b1b
• Personal Stuff: Training AI (Actual Intelligence: dog), breaking game physics, Solar-only model training

🧪 Proof of Concept

import torch

x, y = torch.tensor([[1.0]]), torch.tensor([[5.0]])
model = torch.nn.Linear(1, 1, bias=False)
opt = torch.optim.SGD(model.parameters(), lr=1.0)

loss = torch.nn.functional.mse_loss(model(x), y)
loss.backward(); opt.step()

# Loss: 0.000000