# Progressive Knowledge Distillation Of Stable Diffusion XL Using Layer Level Loss

arXiv: https://arxiv.org/abs/2401.02677 · HTML: https://mlsanity.com/paper/2401.02677/

## Key facts

- Cited by 52 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models.
- Builds on 7 of its 14 references that are in the map.
- First posted on arXiv 5 Jan 2024.

## Placed in

- Diffusion models / [Diffusion foundations and sampling](https://mlsanity.com/diffusion/diffusion-foundations-and-sampling/)

## Builds on (most cited first, 7 of 7)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://mlsanity.com/paper/1910.01108/) (2019)
- [Zero-Shot Text-to-Image Generation](https://mlsanity.com/paper/2102.12092/) (2021)
- [Kosmos-2: Grounding Multimodal Large Language Models to the World](https://mlsanity.com/paper/2306.14824/) (2023)
- [Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference](https://mlsanity.com/paper/2310.04378/) (2023)
- [Wuerstchen: An Efficient Architecture for Large-Scale Text-to-Image Diffusion Models](https://mlsanity.com/paper/2306.00637/) (2023)

## Built on by (most cited first, 30 of 52)

- [FakeInversion: Learning to Detect Images from Unseen Text-to-Image Models by Inverting Stable Diffusion](https://mlsanity.com/paper/2406.08603/) (2024)
- [Faster Diffusion via Temporal Attention Decomposition](https://mlsanity.com/paper/2404.02747/) (2024)
- [LAPTOP-Diff: Layer Pruning and Normalized Distillation for Compressing Diffusion Models](https://mlsanity.com/paper/2404.11098/) (2024)
- [Inference-time Scaling of Diffusion Models through Classical Search](https://mlsanity.com/paper/2505.23614/) (2025)
- [FIRE: Robust Detection of Diffusion-Generated Images via Frequency-Guided Reconstruction Error](https://mlsanity.com/paper/2412.07140/) (2024)
- [CO-SPY: Combining Semantic and Pixel Features to Detect Synthetic Images by AI](https://mlsanity.com/paper/2503.18286/) (2025)
- [A-Bench: Are LMMs Masters at Evaluating AI-generated Images?](https://mlsanity.com/paper/2406.03070/) (2024)
- [EvalMuse-40K: A Reliable and Fine-Grained Benchmark with Comprehensive Human Annotations for Text-to-Image Generation Model Evaluation](https://mlsanity.com/paper/2412.18150/) (2024)
- [AIGIQA-20K: A Large Database for AI-Generated Image Quality Assessment](https://mlsanity.com/paper/2404.03407/) (2024)
- [The ML.ENERGY Benchmark: Toward Automated Inference Energy Measurement and Optimization](https://mlsanity.com/paper/2505.06371/) (2025)
- [MISC: Ultra-low Bitrate Image Semantic Compression Driven by Large Multimodal Model](https://mlsanity.com/paper/2402.16749/) (2024)
- [DiffCut: Catalyzing Zero-Shot Semantic Segmentation with Diffusion Features and Recursive Normalized Cut](https://mlsanity.com/paper/2406.02842/) (2024)
- [SDXS: Real-Time One-Step Latent Diffusion Models with Image Conditions](https://mlsanity.com/paper/2403.16627/) (2024)
- [Large Multi-modality Model Assisted AI-Generated Image Quality Assessment](https://mlsanity.com/paper/2404.17762/) (2024)
- [BlockDance: Reuse Structurally Similar Spatio-Temporal Features to Accelerate Diffusion Transformers](https://mlsanity.com/paper/2503.15927/) (2025)
- [Seg4Diff: Unveiling Open-Vocabulary Segmentation in Text-to-Image Diffusion Transformers](https://mlsanity.com/paper/2509.18096/) (2025)
- [NTIRE 2024 Quality Assessment of AI-Generated Content Challenge](https://mlsanity.com/paper/2404.16687/) (2024)
- [Learning to Sample Effective and Diverse Prompts for Text-to-Image Generation](https://mlsanity.com/paper/2502.11477/) (2025)
- [Reward Fine-Tuning Two-Step Diffusion Models via Learning Differentiable Latent-Space Surrogate Reward](https://mlsanity.com/paper/2411.15247/) (2024)
- [Efficient Fine-Tuning and Concept Suppression for Pruned Diffusion Models](https://mlsanity.com/paper/2412.15341/) (2024)
- [EdgeFusion: On-Device Text-to-Image Generation](https://mlsanity.com/paper/2404.11925/) (2024)
- [Image Quality Assessment: From Human to Machine Preference](https://mlsanity.com/paper/2503.10078/) (2025)
- [Neodragon: Mobile Video Generation using Diffusion Transformer](https://mlsanity.com/paper/2511.06055/) (2025)
- [ComfyGen: Prompt-Adaptive Workflows for Text-to-Image Generation](https://mlsanity.com/paper/2410.01731/) (2024)
- [LoRA-X: Bridging Foundation Models with Training-Free Cross-Model Adaptation](https://mlsanity.com/paper/2501.16559/) (2025)
- [CMC-Bench: Towards a New Paradigm of Visual Signal Compression](https://mlsanity.com/paper/2406.09356/) (2024)
- [DRAGON: A Large-Scale Dataset of Realistic Images Generated by Diffusion Models](https://mlsanity.com/paper/2505.11257/) (2025)
- [Unsupervised Segmentation by Diffusing, Walking and Cutting](https://mlsanity.com/paper/2412.04678/) (2024)
- [DIP: Unsupervised Dense In-Context Post-training of Visual Representations](https://mlsanity.com/paper/2506.18463/) (2025)
- [EC-Diff: Fast and High-Quality Edge-Cloud Collaborative Inference for Diffusion Models](https://mlsanity.com/paper/2507.11980/) (2025)

## Abstract (arXiv)

Stable Diffusion XL (SDXL) has become the best open source text-to-image model (T2I) for its versatility and top-notch image quality. Efficiently addressing the computational demands of SDXL models is crucial for wider reach and applicability. In this work, we introduce two scaled-down variants, Segmind Stable Diffusion (SSD-1B) and Segmind-Vega, with 1.3B and 0.74B parameter UNets, respectively, achieved through progressive removal using layer-level losses focusing on reducing the model size while preserving generative quality. We release these models weights at https://hf.co/Segmind. Our methodology involves the elimination of residual networks and transformer blocks from the U-Net structure of SDXL, resulting in significant reductions in parameters, and latency. Our compact models effectively emulate the original SDXL by capitalizing on transferred knowledge, achieving competitive results against larger multi-billion parameter SDXL. Our work underscores the efficacy of knowledge distillation coupled with layer-level losses in reducing model size while preserving the high-quality generative capabilities of SDXL, thus facilitating more accessible deployment in resource-constrained environments.