丟棄而後恢復：視覺-語言-行動模型有多冗餘？

摘要

視覺-語言-動作（VLA）模型使得基於指令的機器人操作成為可能，但它們繼承了預訓練VLM中過於龐大的語言骨幹，其容量遠超過處理簡短機器人指令所需。這引發了一個基本問題：對於閉環控制而言，VLA模型實際上需要多少部分？在本研究中，我們以Transformer區塊移除作為受控干預手段，探討VLA模型中的架構冗餘。我們提出「捨棄與恢復」（DTR）分析協議：從預訓練VLA模型中移除指定區塊，然後微調所得模型，以衡量被移除的容量是否對下游控制不可或缺。為使此干預可靠，我們提出GateProbe——一種一次性虛擬閘極敏感度指標，用以根據區塊對下游動作損失的貢獻程度進行排名。在多種VLA架構、操作基準測試，甚至真實機器人工業場景中，我們發現移除後的可恢復性呈現強烈不對稱性：**語言骨幹對於標準機器人操作任務高度冗餘，而視覺與動作路徑則對移除的容忍度顯著較低**。在LIBERO上，移除一半LLM區塊甚至能使OpenVLA-OFT在相同的下游微調預算下，從95.0%提升至98.3%；僅保留兩個語言區塊仍能恢復基準級效能。這些結果表明，當前的VLA基準可能對深層語言接地與組合式指令理解施加的壓力有限，未來VLA架構應更審慎地在語言、視覺與動作組件之間分配容量。程式碼已公開於 https://github.com/s1ghhh/VLADrop。

English

Vision-Language-Action (VLA) models enable instruction-driven robotic manipulation, but they inherit oversized language backbones from pretrained VLMs whose capacity far exceeds what is needed for short robotic instructions. This raises a basic question: how much of a VLA model is actually necessary for closed-loop control? In this work, we study architectural redundancy in VLA models by using transformer block removal as a controlled intervention. We introduce Drop-Then-Recovery (DTR), an analysis protocol that removes selected blocks from a pretrained VLA model and then fine-tunes the resulting model to measure whether the removed capacity was necessary for downstream control. To make this intervention reliable, we propose GateProbe, a one-shot virtual-gate sensitivity metric that ranks blocks by their contribution to the downstream action loss. Across multiple VLA architectures, manipulation benchmarks and even real-robot industrial scenarios, we find a strong asymmetry in post-removal recoverability: \textit{language backbones are highly redundant for standard robotic manipulation tasks, whereas vision and action pathways are substantially less tolerant to removal}. On LIBERO, removing half of the LLM blocks even improves OpenVLA-OFT from 95.0% to 98.3% under the same downstream fine-tuning budget, and retaining only two language blocks still recovers baseline-level performance. These results suggest that current VLA benchmarks may exert limited pressure on deep language grounding and compositional instruction understanding, and that future VLA architectures should allocate capacity more deliberately across language, vision, and action components. The code is available at https://github.com/s1ghhh/VLADrop.