GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash Speed
Autoregressive (AR) grounding models serialize spatial predictions, introducing sequential latency and imposing a causal order on output tokens.
ProofPaper ↗
Key points
- We view grounding as visual evidence extraction: objects, locations, and spatial relations are jointly constrained by the image and query, yet their dependencies do not imply an intrinsic left-to-right generation order.
- We introduce GroundAnything, a 4B-parameter grounding foundation model that reconciles fast parallel decoding with precise localization through blockwise denoising.
- With entropy-guided decoding, GroundAnything also surpasses the prior state of the art at this scale, averaging 61.75% versus 53.32% for the fast MTP-based LocateAnything model.
- Infrastructure experiments show that progressive inference optimizations translate parallel decoding into practical speedups.
Sources (1)
- [1]GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash SpeedHugging Face Daily Papers · Sep 30, 12:00 AM
Autoregressive (AR) grounding models serialize spatial predictions, introducing sequential latency and imposing a causal order on output tokens.
We view grounding as visual evidence extraction: objects, locations, and spatial relations are jointly constrained by the image and query, yet their dependencies do not imply an intrinsic left-to-right generation order.
Extractive summary: sentences quoted from the sources.
Before this
- Sep 29, 2026Astra 6.1 Pulled As Insufficiently Aligned
- Sep 29, 2026Sonnet 5.5 is worth a try
- Sep 28, 2026Basis completes a tax workbook 2x faster with GPT-6 Astra
- Sep 23, 2026Show HN: Made an open-source Lego AI generator
- Aug 26, 2026vllm-project/vllm v0.28.0
- Jun 9, 2026Introducing Gemma 4 12B: a unified, encoder-free multimodal model