They tested this with 400,000 GPU hours across multiple RL recipes: DeepSeek (GRPO)
Qwen (DAPO)
Magistral
Minimax Result: only ScaleRL showed a stable, predictable trajectory. The others broke scaling laws entirely their curves collapsed.
COMPUTING
-
ScaleRL outperforms in RL scaling tests
By
–
-
Meta Reveals RL Scaling Challenges
By
–
Today, everyone talks about scaling models. But Meta just proved we’ve been ignoring the harder problem scaling reinforcement learning compute. Turns out, most RL methods don’t scale like pretraining. They plateau early burning millions in compute for almost no gain. ScaleRL
-
Meta’s ScaleRL reveals predictable RL scaling laws
By
–
Holy shit… Meta just cracked the art of scaling RL for LLMs. For the first time ever, they showed that "reinforcement learning follows predictable scaling laws" just like pretraining. Their new framework, 'ScaleRL', fits a sigmoid compute-performance curve that can forecast
-
Colleague regrets Intel upgrade before M1 release
By
–
之前同事刚好到年限换机器，换了 16 的 intel，结果半年后 m1 就出了，我们都是 m1，他特别难受。
-
GPU Computing: The New Gold Rush Replacing SaaS Software
By
–
if you’re in software, pivot to GPUs now take on debt, sell a kidney, whatever
but get your hands on that silicon the SaaS gold rush is over
new money is compute the next king isn’t code
it’s hardware -
Global Leaders Building Next-Generation AI Infrastructure at NVIDIA GTC
By
–
We're excited to these discussions on how global leaders are building their next-generation AI infrastructure at #NVIDIAGTC Washington, D.C.
-
GPU PCIe Layout Mapping Guide for AI Builders
By
–
> this is how you map out a 4x GPU PCIe layout
> every builder needs to know this cold
> Buy a GPU is almost here
> and it’s lane-checking all your builds —Buy a GPU, The Movement -
Commonwealth Fusion Systems ARC to Grid Early 2030s
By
–
Going off of public statements, CFS’s ARC is expected to deliver power to the grid in the early 2030s:
-
Token Ring Network Architecture and Performance Analysis
By
–
token ring , il tourne sûrement en rond
-
NVIDIA Delivers DGX Spark Supercomputer to Elon Musk at Starbase
By
–
From rockets to AI. Nine years after the original NVIDIA DGX-1 handoff, Jensen Huang delivered a brand-new DGX Spark to @ElonMusk at Starbase. The DGX Spark is a petaflop powerhouse built for creators, researchers, and developers — a desktop-sized supercomputer with five