Meta’s Generative Ads Recommendation Model (GEM), the engine behind ads on Instagram and Facebook, is now training at LLM scale across thousands of the latest-gen GPUs.
Because standard LLM infrastructure doesn't natively fit recommendation workloads, we co-designed custom kernels like Jagged Flash Attention (JFA), Generalized Dot-Product Attention (GDPA), BlockAttention, ultra-low mixed precision (including MXFP8 attention), and a topology-aware 5D parallelism architecture tailored to our network hierarchy.
The result: we scaled total training FLOPs 4x in 12 months while doubling end-to-end training efficiency to 20–25% MFU.
Read the full deep dive here: https://t.co/btMBljsNVS
Unrelated but this guy made his money buying up all the innovative 90s educational software companies (creators of Reader Rabbit, Oregon Trail, Carmen Sandiego, et al), completely gutting them, and then selling their husks to Mattel.
He singlehandedly destroyed that sector.
Pat Tillman’s communications were monitored and sent up the chain from the day he enlisted. Once they saw he had turned against the war and planned to publicly oppose it on his return a decision was made to kill him. They weren’t going to risk another Smedley Butler
Log volumes are growing. Your costs shouldn't be.
Amazon OpenSearch Service now delivers up to 4X better price performance for log analytics, with search and analytics unified in one service.
Sonnet 5 provides big gains over 4.6 across reasoning, tool use, coding, and knowledge work.
It's built for real engineering work. Navigating large codebases, landing multi-file changes, and carrying long debugging or refactor tasks through to completion.
Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor.
It’s happening faster than we thought, and the implications deserve greater attention. https://t.co/OVVPJO7VQx
"The vocation of the learner in the age of cheap wheat is to become a baker: to take the now-abundant raw material and turn it into something a human can eat."
NASA FIRMS satellite detects MULTIPLE FIRES across Hormuz following tonight's US-Iran exchange of ATTACKS
'LIKELY DRIFTING FIRE' signatures consistent with possible 'BURNING VESSELS'
Usage limits are up, effective today we're:
1) Doubling Claude Code's 5-hour limits for Pro, Max, Team and seat-based Enterprise plans
2) Removing peak hours limit reduction on Claude Code for Pro and Max plans
3) Substantially raising our API rate limits for Opus models
Introducing Claude Opus 4.7, our most capable Opus model yet.
It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.
You can hand off your hardest work with less supervision.