Dillon Cower's Avatar

Dillon Cower

@dcower.bsky.social

engineer @ waymo, xoogler foundation models & all things data 🎸 β†’ @ratan.bsky.social

166 Followers  |  825 Following  |  9 Posts  |  Joined: 20.10.2024  |  1.6253

Latest posts by dcower.bsky.social on Bluesky

yep

20.05.2025 17:38 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

Introducing The AI CUDA Engineer: An agentic AI system that automates the production of highly optimized CUDA kernels.

sakana.ai/ai-cuda-engi...

The AI CUDA Engineer can produce highly optimized CUDA kernels, reaching 10-100x speedup over common machine learning operations in PyTorch.

Examples:

20.02.2025 01:50 β€” πŸ‘ 89    πŸ” 17    πŸ’¬ 3    πŸ“Œ 4

lgtm

11.01.2025 03:48 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

I kind of like "deep thought models" (optional 🚬) -- extra time spent on thinking etc. And it's still generic enough to soon become overloaded! ++nerd reference too.

08.01.2025 02:25 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 1    πŸ“Œ 1

Self-reasoning models? Long-thinking/long-thought models?

08.01.2025 02:19 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

⚰️

22.12.2024 18:08 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

1/ Okay, one thing that has been revealed to me from the replies to this is that many people don't know (or refuse to recognize) the following fact:

The unts in ANN are actually not a terrible approximation of how real neurons work!

A tiny 🧡.

πŸ§ πŸ“ˆ #NeuroAI #MLSky

16.12.2024 20:03 β€” πŸ‘ 151    πŸ” 38    πŸ’¬ 21    πŸ“Œ 17
Video thumbnail

ImPlot3D: A 3D Plotting Library for Dear ImGui
github.com/brenocq/impl...

17.12.2024 21:36 β€” πŸ‘ 151    πŸ” 25    πŸ’¬ 6    πŸ“Œ 0

Computer "Science" should look at real Science. :)
Pre-registering any experiments/trials before research. (E.g., pre-register your prompts/examples!). Not showing a table with a PSNR of 36.142 but confidence intervals and a distribution of N training runs w different hyperparameters/seeds.

11.12.2024 21:59 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

strong indicators of a seattle area alphabet employee's screen: luum bookmark, ever-present colab warnings.

I did not even think of taking Amtrak to get here!

12.12.2024 04:11 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

I'm excited to share a new paper: "Mastering Board Games by External and Internal Planning with Language Models"

storage.googleapis.com/deepmind-med...

(also soon to be up on Arxiv, once it's been processed there)

05.12.2024 07:49 β€” πŸ‘ 76    πŸ” 13    πŸ’¬ 4    πŸ“Œ 7
Post image

A common question nowadays: Which is better, diffusion or flow matching? πŸ€”

Our answer: They’re two sides of the same coin. We wrote a blog post to show how diffusion models and Gaussian flow matching are equivalent. That’s great: It means you can use them interchangeably.

02.12.2024 18:45 β€” πŸ‘ 254    πŸ” 58    πŸ’¬ 6    πŸ“Œ 7

In addition to the Deep Learning Theory starter pack, I've also put together a starter pack for Reinforcement Learning Theory. Let me know if you'd like to be included or suggest someone to add to the list!

go.bsky.app/LWyGAAu

22.11.2024 21:56 β€” πŸ‘ 29    πŸ” 10    πŸ’¬ 11    πŸ“Œ 1

great twet over here

21.10.2024 03:52 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

no. no? no.

20.10.2024 19:41 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

egg

20.10.2024 04:41 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

@dcower is following 19 prominent accounts