Poly-OPD：用於能力可選擇流模型的異構多教師在策略蒸餾

摘要

領先的開放式文生圖模型往往具有互補優勢：一個可能在偏好對齊的美感上領先，另一個則更忠實地遵循構圖指令。然而，它們的自編碼器與噪聲調度之間的差異，使得這些優勢難以在不同模型間遷移。在本文中，我們提出 Poly-OPD，一個能將異質教師模型的互補優勢整合至單一緊湊的流匹配學生模型中的框架。為橋接不同教師之間不相容的潛在空間，Poly-OPD 透過像素橋接進行同策略蒸餾。每張學生生成的影像都會由所選教師的編碼器重新編碼，並在教師的噪聲調度下，從幅度匹配的噪聲水平開始細化。所得目標進一步在凍結的 DINOv2 空間中與學生匹配，從而在不相容的潛在空間之間實現監督。為在無跨教師干擾的情況下保留互補能力，Poly-OPD 使用梯度相容性診斷來組織其適配器：注意力 LoRA 模組在教師間共享，而前饋適配器則保持教師特定。在蒸餾過程中，差距感知的課程會將更多訓練分配給學生仍不如教師的構圖類別；隨著每個差距縮小，訓練會轉向剩餘差距較大的類別。透過將 FLUX.1-dev 和 Z-Image 蒸餾至一個 2.5B 的 SD3.5-Medium 學生模型，Poly-OPD 將 GenEval 從 67.3 提升至 73.3，超越兩個更大的教師模型，並將 DrawBench HPSv3 從 9.34 提高至 11.35，將兩者的優勢整合於一個可切換的模型中。

English

Leading open text-to-image models often carry complementary strengths: one may lead on preference-aligned aesthetics while another follows compositional instructions more faithfully. However, differences in their autoencoders and noise schedules make it difficult to transfer these strengths across models. In this paper, we present Poly-OPD, a framework that can consolidate complementary strengths of heterogeneous teachers into a single compact flow-matching student. To bridge the incompatible latent spaces of different teachers, Poly-OPD performs on-policy distillation through a pixel bridge. Each student-generated image is re-encoded by a selected teacher's encoder and refined from a noise level matched by magnitude under the teacher's noise schedule. The resulting target is further matched to the student in frozen DINOv2 space, enabling supervision across incompatible latent spaces. To retain complementary capabilities without cross-teacher interference, Poly-OPD uses a gradient compatibility diagnostic to organize its adapters: attention LoRA modules are shared across teachers, whereas feed-forward adapters remain teacher-specific. During distillation, a gap-aware curriculum devotes more training to compositional categories where the student still falls short of the teacher. As each gap narrows, training shifts toward categories with larger remaining gaps. By distilling FLUX.1-dev and Z-Image into a 2.5B SD3.5-Medium student, Poly-OPD improves GenEval from 67.3 to 73.3, surpassing both larger teachers, and raises DrawBench HPSv3 from 9.34 to 11.35, consolidating both strengths within a switchable model.