Self-Rewarding Vision-Language Model via Reasoning Decomposition
@_akhaliq
-
Nvidia Presents Autoregressive Universal Video Segmentation Model
By
–
Nvidia presents Autoregressive Universal Segmentation Model
-
Pixie: Fast Supervised 3D Physics Learning from Pixels
By
–
Pixie
— AK (@_akhaliq) 27 août 2025
Fast and Generalizable Supervised Learning of 3D Physics from Pixels pic.twitter.com/7HUIgGyc3SPixie Fast and Generalizable Supervised Learning of 3D Physics from Pixels
-
OmniHuman-1.5: Cognitive Simulation for Active Avatar Minds
By
–
OmniHuman-1.5
— AK (@_akhaliq) 27 août 2025
Instilling an Active Mind in Avatars via Cognitive Simulation pic.twitter.com/OOOyNyc6BQOmniHuman-1.5 Instilling an Active Mind in Avatars via Cognitive Simulation
-
VoxHammer: Training-Free 3D Editing in Native Space
By
–
VoxHammer
— AK (@_akhaliq) 27 août 2025
Training-Free Precise and Coherent 3D Editing in Native 3D Space pic.twitter.com/NkGeimhOdkVoxHammer Training-Free Precise and Coherent 3D Editing in Native 3D Space
-
Distilled 3D Gaussian Splatting Technology
By
–
Distilled-3DGS
— AK (@_akhaliq) 25 août 2025
Distilled 3D Gaussian Splatting pic.twitter.com/MmLgivksRRDistilled-3DGS Distilled 3D Gaussian Splatting
-
Wan2.2-T2V-A14B becomes default text to video model in Anycoder
By
–
Wan2.2-T2V-A14B @Alibaba_Wan is now the default text to video model in anycoder pic.twitter.com/qeBfflKoad
— AK (@_akhaliq) 25 août 2025Wan2.2-T2V-A14B @Alibaba_Wan is now the default text to video model in anycoder
-
Evaluating LLMs’ Performance in Premier Programming Competitions
By
–
AetherCode Evaluating LLMs' Ability to Win In Premier Programming Competitions
-
CRISP: Persistent Concept Unlearning via Sparse Autoencoders
By
–
CRISP Persistent Concept Unlearning via Sparse Autoencoders
-
Self-Play and Variational Problem Synthesis Enhance RLVR Performance
By
–
Beyond Pass@1 Self-Play with Variational Problem Synthesis Sustains RLVR