# SeqLoRA: Bilevel Orthogonal Adaptation for Continual Multi-Concept Generation

arXiv: https://arxiv.org/abs/2605.22743 · HTML: https://mlsanity.com/paper/2605.22743/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Image generation, text-to-image and editing, one of 14 directions in Diffusion models.
- Builds on 25 of its 37 references that are in the map.
- First posted on arXiv 21 May 2026.

## Placed in

- Diffusion models / [Image generation, text-to-image and editing](https://mlsanity.com/diffusion/image-generation-text-to-image-and-editing/)

## Builds on (most cited first, 25 of 25)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://mlsanity.com/paper/2204.06125/) (2022)
- [Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding](https://mlsanity.com/paper/2205.11487/) (2022)
- [SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis](https://mlsanity.com/paper/2307.01952/) (2023)
- [DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation](https://mlsanity.com/paper/2208.12242/) (2022)
- [An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion](https://mlsanity.com/paper/2208.01618/) (2022)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [Multi-Concept Customization of Text-to-Image Diffusion](https://mlsanity.com/paper/2212.04488/) (2022)
- [Human Preference Score v2: A Solid Benchmark for Evaluating Human Preferences of Text-to-Image Synthesis](https://mlsanity.com/paper/2306.09341/) (2023)
- [Spectrally-normalized margin bounds for neural networks](https://mlsanity.com/paper/1706.08498/) (2017)
- [DreamSim: Learning New Dimensions of Human Visual Similarity using Synthetic Data](https://mlsanity.com/paper/2306.09344/) (2023)
- [A PAC-Bayesian Approach to Spectrally-Normalized Margin Bounds for Neural Networks](https://mlsanity.com/paper/1707.09564/) (2017)
- [Mix-of-Show: Decentralized Low-Rank Adaptation for Multi-Concept Customization of Diffusion Models](https://mlsanity.com/paper/2305.18292/) (2023)
- [Break-A-Scene: Extracting Multiple Concepts from a Single Image](https://mlsanity.com/paper/2305.16311/) (2023)
- [HPSv3: Towards Wide-Spectrum Human Preference Score](https://mlsanity.com/paper/2508.03789/) (2025)
- [InfLoRA: Interference-Free Low-Rank Adaptation for Continual Learning](https://mlsanity.com/paper/2404.00228/) (2024)
- [On the Convergence Theory of Gradient-Based Model-Agnostic Meta-Learning Algorithms](https://mlsanity.com/paper/1908.10400/) (2019)
- [State of the Art on Diffusion Models for Visual Computing](https://mlsanity.com/paper/2310.07204/) (2023)
- [Orthogonal Adaptation for Modular Customization of Diffusion Models](https://mlsanity.com/paper/2312.02432/) (2023)
- [LoRACLR: Contrastive Adaptation for Customization of Diffusion Models](https://mlsanity.com/paper/2412.09622/) (2024)
- [Taming Diffusion Models for Image Restoration: A Review](https://mlsanity.com/paper/2409.10353/) (2024)

## Built on by (most cited first, 1 of 1)

- [CLASP: Continual Low-rank Adapters for Spatially Placed Concepts from One Hypernetwork](https://mlsanity.com/paper/2610.01331/) (2026)

## Abstract (arXiv)

Parameter-efficient fine-tuning enables fast personalization of text-to-image diffusion models to user-provided concepts (objects, people, or styles), but composing multiple such concepts remains challenging due to representation interference. Existing modular methods, usually built on low-rank adaptation (LoRA), either rely on expensive post-hoc fusion or freeze the LoRA adaptation subspaces, which limit expressiveness and concept fidelity. To address this trade-off, we propose Sequential regularized LoRA (SeqLoRA), a constrained continual learning framework that jointly optimizes both LoRA factors via bilevel optimization while keeping each new basis orthogonal to all previously learned ones. Theoretically, we establish monotone descent and convergence to a critical point of the constrained problem, and model the residual layer activations as a matrix sub-Gaussian process to derive a high-probability bound on catastrophic forgetting in multi-layer nonlinear networks. This bound depends on the basis only through a residual interference energy, and within the feasible subspace we prove that a data-adapted basis minimizes it, whereas a random frozen basis is suboptimal in expectation. Experiments on Stable Diffusion with up to 101 concepts show that SeqLoRA improves identity preservation over fusion-based methods, attains the lowest cross-concept leakage in multi-concept compositions, requires no fusion step, and scales to concept counts at which fusion runs out of memory.