Averaging and temporally interpolating text latents in VLAs enables 83% success on novel task combinations in the libero-ood benchmark where SOTA models achieve under 15%.
Sigmoid loss for language image pre-training
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
citation-role summary
method 1
citation-polarity summary
roles
method 1polarities
use method 1representative citing papers
HoloCount is a three-tier visual counting benchmark showing that MLLMs fail systematically on analytical reasoning, high-density scenes, and linguistic prior conflicts, with even the best models dropping below 50% accuracy on dense counting.
Omni-Encoder unifies visual and audio encoding at symmetrical 25 fps using a Transformer with three new components, yielding gains on fine-grained motion tasks while matching baselines on audio-visual benchmarks.
ALLaVA creates 1.3M GPT4V-synthesized samples enabling 4B VLMs to achieve competitive results on 17 benchmarks and match 7B/13B models on some tasks.
Show-o2 unifies text, image, and video understanding and generation in a single autoregressive-plus-flow-matching model built on 3D causal VAE representations.
citing papers explorer
-
VLAs are Confined yet Capable of Generalizing to Novel Instructions
Averaging and temporally interpolating text latents in VLAs enables 83% success on novel task combinations in the libero-ood benchmark where SOTA models achieve under 15%.
-
HoloCount: A Holistic Visual Counting Benchmark for MLLMs
HoloCount is a three-tier visual counting benchmark showing that MLLMs fail systematically on analytical reasoning, high-density scenes, and linguistic prior conflicts, with even the best models dropping below 50% accuracy on dense counting.
-
OmniEncoder: See, Hear, and Feel Continuous Motion Like Humans With One Encoder
Omni-Encoder unifies visual and audio encoding at symmetrical 25 fps using a Transformer with three new components, yielding gains on fine-grained motion tasks while matching baselines on audio-visual benchmarks.
-
ALLaVA: Harnessing GPT4V-Synthesized Data for Lite Vision-Language Models
ALLaVA creates 1.3M GPT4V-synthesized samples enabling 4B VLMs to achieve competitive results on 17 benchmarks and match 7B/13B models on some tasks.
-
Show-o2: Improved Native Unified Multimodal Models
Show-o2 unifies text, image, and video understanding and generation in a single autoregressive-plus-flow-matching model built on 3D causal VAE representations.