Latest in AI and developer tools

Today's trail

updated 4m ago

Pulse · msn.com

msn.com · Nov 13

Nvidia's rivals, from AMD to Google, and the chips they are betting on to loosen its AI grip

Nvidia supplies about three-fifths of AI computing power. How AMD, Google, Amazon and Huawei compete, and why gains may come slowly.

Pulse · hindustantimes.com

hindustantimes.com · Nov 13

Nvidia's $10 billion bet on Anthropic IPO, and the chip guarantees behind the AI industry's borrowing

Jensen Huang's exit as Nvidia's chief executive, a week before Anthropic is due to begin trading in what could be the biggest stock market listing on record, has put a spotlight on the chipmaker's $10 billion pledge to the artificial intelligence (AI ...

Pulse

Top picks this weekarXiv

arXiv

Nov 12

Back to Pixels: Lifting Latent Video Diffusion into Pixel Space

Frontier video generation models work in a VAE latent space that compresses video in both space and time. This compression made training tractable, but now limits high-resolution, controllable, and interactive generation -- doubling resolution quadruples token count, conditions apply to groups of frames rather than individual ones, and each response must wait for the rest of its multi-frame block. We present PixVid, a simple yet effective method that lifts a pretrained latent video model to directly generate every frame in pixel space. PixVid keeps the diffusion transformer and replaces the VAE decoder with a convolutional head that renders its outputs into pixels. We find that predicting clean video rather than velocity, together with a larger noise shift, is key to quality. Finetuned from Wan2.1-14B, PixVid is competitive with the latent baseline within $\sim 1.3$M training clips. Working framewise in pixels, it generates at $2\times$ higher resolution with the same per-frame token count, keeps depth-conditioned video editing aligned under fast motion, and responds to each action at the next frame. Project page: https://chenshuo20.github.io/PixVid/

Top picks

All arXiv Hugging Face GitHub Hacker News

Latest Top this week

arXiv

Sep 21

Extracting Arguments, Not Just Classifying Them: Instruction-Tuned LLMs for Generative Component Detection

AI Instruction-tuned LLMs for generative argument component detection

arXiv

Sep 21

Revisiting Multi-View Stereo: A Sequence-to-Sequence Formulation

Computing accurate geometry from multi-view images is a fundamental problem in computer vision. Recent feed-forward (FF) models jointly estimate 3D geometry and camera parameters, but they typically suffer from geometry distortion caused by reconstruction ambiguity, even when…

arXiv

Sep 21

SPECTRA: Adaptive Execution of Speculative Decoding on a Runtime-Reconfigurable Tiled Architecture

AI Runtime-reconfigurable architecture for speculative decoding.

arXiv

Sep 21

PredActor: Predictive Action Diffusion for Steerable Onboard Humanoid Control

AI Diffusion-based predictive action policy for humanoid robot control

This week in Multimodal

This week

Get the morning digest

Today's trail

Pulse

Top picks

Today's trail

Rising tools

This week in Multimodal

Most active this week