Yang Wang

A unified policy enables humanoid robots to robustly interact with complex environments, achieving record-breaking success rates in dynamic terrain and command scenarios.

Source-style collapse can cause fine-tuned retrievers to miss relevant capabilities, but a simple TF-IDF signal can dramatically improve retrieval success rates.

GigaBrain-0.7 achieves unprecedented task adaptability and completion rates, outperforming previous models in both home and industrial settings.

WidgetGen achieves superior visual reconstruction performance without the constraints of fixed UI schemas, redefining the widget-to-code generation landscape.

A simple screenshot-based judge often outperforms complex LLM evaluation methods, challenging assumptions about the necessity of intricate judging pipelines.

Current video-language models struggle to interpret the nuanced meanings behind social media videos, often missing the implicit narratives that make them humorous or ironic.

CI-Diff successfully decouples unusual attributes from common associations, enabling accurate image synthesis for rare concepts that previous models struggled to render.

Action-only decoding in GigaWorld-Policy-0.5 slashes inference latency to 85 ms, revolutionizing real-time robot control efficiency.

Breaking the tight coupling between feature engineering and model architecture, PG enables rapid iterations and deployment in generative retrieval systems, yielding measurable performance gains.

Evaluator quality for robotic policies hinges more on long-horizon consistency than on short-term visual fidelity, reshaping our approach to world model design.

Agentic models can learn to trust their "gut" and rely less on external tools, leading to faster and more accurate reasoning.

Robots can now better assemble boxes in the real world thanks to a video-generative value model that anticipates future states, moving beyond static snapshots for more reliable task progress assessment.

LLMs reason better when their uncertainty consistently decreases, paving the way for shorter, more accurate chain-of-thought reasoning.

Forget mixed-precision: tunable INT8 emulation can simultaneously boost accuracy and performance in FP64 HPC workloads on GPUs.

Robots can now plan 9x faster and achieve significantly higher success rates by decoupling action prediction from video generation in World-Action Models.

Ditch the handcrafted coefficients: DyWeight learns how to dynamically weight gradients in diffusion model sampling, slashing compute while boosting image quality.

Forget end-to-end VLAs: GigaBrain-0.5M* leverages world models and reinforcement learning to achieve a 30% performance boost on complex robotic manipulation tasks, showcasing reliable long-horizon execution.