菜单
Bridge the modality gap between language reasoning and low-level robot actions through two-stage alignment training. Use RL with action-accuracy rewards to ground linguistic reasoning in precise motor control, achieving 17.5%-106.6% improvements over SFT baselines.
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 | roboalign-language-action-reasoning |
| title | RoboAlign: Learning Test-Time Reasoning for Language Action Alignment in Vision Language Action Models |
| version | 0.0.3 |
| engine | skillxiv-v0.0.3-claude-opus-4.6 |
| license | MIT |
| url | https://arxiv.org/abs/2603.21341 |
| keywords | ["VLA","Embodied AI","Language-Action Alignment","Test-Time Reasoning","Robot Control"] |
| description | Bridge the modality gap between language reasoning and low-level robot actions through two-stage alignment training. Use RL with action-accuracy rewards to ground linguistic reasoning in precise motor control, achieving 17.5%-106.6% improvements over SFT baselines. |
Two-Stage Language-Action Alignment Framework for VLAs
Multimodal large language models (MLLMs) excel at embodied reasoning tasks, but these improvements don't consistently translate to better robot control. The gap exists because "modality gap between language and low-level actions"—language understanding and motor control are optimized separately. Reasoning improvements don't guarantee action accuracy.
RoboAlign introduces two-stage training that directly aligns MLLM capabilities with robot action generation:
Stage 1: Supervised Fine-Tuning (SFT) - Train the model to generate low-level action tokens through reasoning. The model learns to emit action tokens as direct outcomes of spatial and temporal reasoning over visual inputs.
Stage 2: Reinforcement Learning (RL) - Apply RL with action-accuracy rewards to refine reasoning toward precise action execution. Instead of optimizing reasoning through language alone, sample action tokens as direct outcomes and evaluate reasoning quality through action accuracy feedback.
This grounds linguistic understanding in concrete motor control, bridging the modality gap.
The mechanism operates by:
The key insight is coupling language reasoning directly to motor control outcomes rather than optimizing them independently.
The paper demonstrates substantial gains:
