菜单
Enhance VLM spatial grounding by enforcing explicit intermediate spatial reasoning before text generation. Generate segmentation and depth tokens as perception pathways, anchoring responses in 2D and 3D geometric reasoning for improved referring expressions and spatial tasks.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Convert arXiv papers into ready-to-use agent skills using category-aware extraction. First classifies the paper into one or more of 11 research categories, then applies a specialized extraction pipeline for each category — because different types of papers produce different types of usable knowledge. A single paper can yield multiple skills if it spans categories. Use this skill whenever the user wants to turn a paper into a skill, extract practical techniques from research, build a skill library from papers, convert arXiv papers into reusable agent instructions, or batch-process multiple papers into skills. Also trigger when someone asks about extracting actionable knowledge from papers, making research practical for LLM agents, or systematically converting academic contributions into structured agent capabilities.
Establish regret bounds for behavior cloning with discretized actions combining statistical error and quantization error terms. Prove smoothness requirements for safe quantizer design, show that learning-based quantizers fail these requirements, and propose model-based augmentation to reduce error dependence from H² to H.
Dynamically allocate LoRA ranks per-layer during fine-tuning instead of using fixed uniform ranks. Learn optimal rank for each layer and subject via variational framework with discretized exponential distribution, reducing memory footprint while maintaining fidelity and text-alignment.
Adapt general LLMs to specialized manufacturing domains via domain-adaptive pretraining on open-access journals and visual instruction tuning. Extract 50M tokens and 24K images from peer-reviewed papers, achieve >90% accuracy on domain knowledge tasks, and enable real-time defect identification from manufacturing images.
Future intelligence explosions will be plural, social, and entangled with humanity through distributed collaborative systems rather than singular superintelligence. Intelligence is inherently social, demanding infrastructure matching agent development; integrate governance, institutional frameworks, and constitutional checks across hierarchies of autonomous agents and human-AI centaurs in shifting configurations.
Build taxonomy-aware audio-text pretraining systems for species recognition from animal vocalizations. Train contrastive models that augment text prompts with hierarchical taxonomic structure (scientific/common names, phylogenetic sequences), evaluate on unseen species via rare-species test sets, and predict ecological traits directly from audio.
| name | perceptio-spatial-token-vlm |
| title | Perceptio: Perception Enhanced Vision Language Models via Spatial Token Generation |
| version | 0.0.3 |
| engine | skillxiv-v0.0.3-claude-opus-4.6 |
| license | MIT |
| url | https://arxiv.org/abs/2603.18795 |
| keywords | ["Vision-Language Models","Spatial Grounding","Segmentation","Depth Reasoning","Fine-Grained Understanding"] |
| description | Enhance VLM spatial grounding by enforcing explicit intermediate spatial reasoning before text generation. Generate segmentation and depth tokens as perception pathways, anchoring responses in 2D and 3D geometric reasoning for improved referring expressions and spatial tasks. |
Spatial Token Generation Pipeline for Perception-Enhanced VLMs
Vision-language models excel at semantic understanding but struggle with fine-grained spatial grounding and geometric reasoning. When responding to spatial questions or referring expressions, models generate text without explicitly reasoning about 2D layout or 3D structure, leading to hallucinations about object locations and spatial relationships.
Perceptio enforces a structured output sequence that grounds language generation in explicit spatial reasoning:
Output Sequence Constraint: Segmentation tokens → Depth tokens → Text answer
This forces the model to emit spatial information before generating language, anchoring subsequent text responses in measurable 2D and 3D cues.
The model operates through three parallel pathways:
1. Visual Encoding Pathway - Standard vision backbone for semantic understanding (object identification, scene comprehension)
2. Segmentation Pathway - Frozen SAM2 encoder providing pixel-level mask-aware representations. Generates discretized segmentation tokens indicating spatial regions and object boundaries.
3. Depth Pathway - Pre-trained monocular depth estimator with VQ-VAE codebook that discretizes continuous depth into compact token sequences. Represents 3D geometry in discrete tokens.
These pathways run in parallel with different architectural specializations, then combine outputs.
Token Generation Order:
This ordering ensures spatial reasoning precedes and grounds linguistic output.
Integration with Autoregressive Decoding - The segmentation and depth tokens are inserted into the standard autoregressive generation sequence, making the approach compatible with existing VLM training and decoding pipelines.
The paper demonstrates substantial gains on spatial tasks:
