Loading activity…
Amdahl comes for us all mts & co-founder @coreautoai @pytorch @gpu_mode enjoyer
AI
Core Automation describes two sites with different GPUs and parallelism schemes connected by a bidirectional 25 Gbps link.
AI
The contest page lays out GPU-kernel and SGLang serving work, with different tests for each.
AI
A post notes that companies want to vertically integrate AI software factory stacks while developer tools and model labs compete.
AI
The organizer says they’re throwing it with two others and shares a joke about hosting it in San Jose.
AI
SemiAnalysis says an upstream vLLM optimization made the 2.8-trillion-parameter model 1.68 times faster on the same hardware.
AI
A poster jokes that Matej's draft claims a Coreauto member-of-technical-staff role focused on reinforcement-learning performance.
AI
The person inviting applications says the work would involve collaborating closely with them.
AI
The author said the retrospective was taking a while, quipping that “it’s not about auto research haha.”
AI
Core Automation describes what worked across 15,602 submissions to the competition.
AI
The user describes the shared image as “the perfect encapsulation” of AI progress over the six months leading up to September 11, 2026.
AI
Company presents the harness as token-efficient for autonomous coding work.
AI
Mark Saroufim posted Tilde Research's blog on models that resist learning and noted the end of its beta competition.
AI
Mark Saroufim posted a video timeline of talks and shared photos from the event.
XAI
Thread solicits single most influential papers across hardware, systems, and ML topics.
AI
Tech leaders and firms endorse open-weight AI models for safety and innovation.
