🏢 University of Washington

Open Deep Search (ODS): Democratizing Search with Open-source Reasoning Agents.

Sync your moves! MusicInfuser adapts video diffusion to make models listen and dance to music, preserving style and aligning movement.

RDS+ is the unsung hero for scaling instruction tuning data selection!

Small language models struggle to learn complex reasoning from large models, but a novel ‘Mix Distillation’ method balances complexity for effective capability transfer.

LLMs struggle with complex logical reasoning; ZebraLogic benchmark reveals a ‘curse of complexity’, highlighting inherent limitations and guiding future research.

BLT: tokenizer-free LLM for efficiency and robustness

Boosting visual reasoning in multimodal language models, AURORA leverages novel ‘Perception Tokens’ for improved depth estimation and object counting.

NegToMe: Image-based adversarial guidance improves image generation diversity and reduces similarity to copyrighted content without training, simply by using images instead of negative text prompts.

GMAI-VL-5.5M & GMAI-VL: A new multimodal medical dataset and vision-language model achieve state-of-the-art results in various medical tasks.

SAMURAI enhances the Segment Anything Model 2 for real-time, zero-shot visual object tracking by incorporating motion-aware memory and motion modeling, significantly improving accuracy and robustness.

Larger language models aren’t always better teachers for instruction tuning; a new metric, CAR, predicts teacher model effectiveness better than existing methods.