Pith. Sign in

RECORD

DiffNAS: Bootstrapping Diffusion Models by Prompting for Better Architectures

PaperPDF from arXiv

data files

Paper breakdownThe authors' claims, conditions and measurements, as JSON Signed receiptProof this review has not been altered
X Bluesky LinkedIn Reddit Hacker News
Wenhao Li, Xiu Su, Shan You, Fei Wang, Chen Qian, Chang Xu

Artificial Intelligence · Submitted 7 Oct 2023

abstract compiled · 2026-09-23 · no language-model review

The authors' main claim Rigorous experimentation corroborates that our algorithm can augment the search efficiency by 2 times under GPT-based scenarios, while also attaining a performance of 2.82 with 0.37 improvement in FID on CIFAR10 relative to the benchmark IDDPM algorithm.

how it is stated · asserted 1.00 read the record →

Comment (0) Read on arXiv ↗
Cite ▾
arXiv ID2310.04750 (version 2)
Pith IDOAWT3KRO, Pith's permanent record of this paper and its review
CitationChang Xu, Chen Qian, Fei Wang, Shan You, Wenhao Li, Xiu Su. "DiffNAS: Bootstrapping Diffusion Models by Prompting for Better Architectures." arXiv:2310.04750 (2023). https://pith.science/paper/OAWT3KRO
Claim paper

The compiled record

The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Open the compiled record

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 6,219 input tokens.

What the abstract itself states, as probabilities: claims first 0.05 · states limitation 0.09 · human participants 0.03 · controlled comparison 0.91 · randomized assignment 0.02 · states quantitative result 0.99 · compares with prior results 0.97 · mentions code or data availability 0.03.

The abstract, sentence by sentence

arXiv listing abstract v2. Results and the main claim are underlined; .

Diffusion models have recently exhibited remarkable performance on synthetic data. After a diffusion path is selected, a base model, such as UNet, operates as a denoising autoencoder, primarily predicting noises that need to be eliminated step by step. Consequently, it is crucial to employ a model that aligns with the expected budgets to facilitate superior synthetic performance. In this paper, we meticulously analyze the diffusion model and engineer a base model search approach, denoted "DiffNAS". Specifically, we leverage GPT-4 as a supernet to expedite the search, supplemented with a search memory to enhance the results. Moreover, we employ RFID as a proxy to promptly rank the experimental outcomes produced by GPT-4. We also adopt a rapid-convergence training strategy to boost search efficiency. Rigorous experimentation corroborates that our algorithm can augment the search efficiency by 2 times under GPT-based scenarios, while also attaining a performance of 2.82 with 0.37 improvement in FID on CIFAR10 relative to the benchmark IDDPM algorithm.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#sentencerolethe authors' own workstates the main conclusionstrength
s0 Diffusion models have recently exhibited remarkable performance on synthetic data. background 1.00 0.04 0.02 asserted 0.99
s1 After a diffusion path is selected, a base model, such as UNet, operates as a denoising autoencoder, primarily predicting noises that need to be eliminated step by step. background 0.98 0.06 0.02 asserted 1.00
s2 Consequently, it is crucial to employ a model that aligns with the expected budgets to facilitate superior synthetic performance. background 0.85 0.07 0.04 asserted 0.98
s3 In this paper, we meticulously analyze the diffusion model and engineer a base model search approach, denoted "DiffNAS". contribution 0.99 0.95 0.05 asserted 1.00
s4 Specifically, we leverage GPT-4 as a supernet to expedite the search, supplemented with a search memory to enhance the results. method 0.97 0.93 0.04 asserted 1.00
s5 Moreover, we employ RFID as a proxy to promptly rank the experimental outcomes produced by GPT-4. method 0.98 0.93 0.03 asserted 0.99
s6 We also adopt a rapid-convergence training strategy to boost search efficiency. method 0.96 0.91 0.03 asserted 0.97
s7 Rigorous experimentation corroborates that our algorithm can augment the search efficiency by 2 times under GPT-based scenarios, while also attaining a performance of 2.82 with 0.37 improvement in FID on CIFAR10 relative to the benchmark IDDPM algorithm. result 1.00 0.96 0.83 asserted 1.00

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.95.

What was examined, and what was not

sources read
  • arXiv listing abstract v2
not examined
  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers
levels run
  • P abstract
not run
  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

The paper
Original abstract

Diffusion models have recently exhibited remarkable performance on synthetic data. After a diffusion path is selected, a base model, such as UNet, operates as a denoising autoencoder, primarily predicting noises that need to be eliminated step by step. Consequently, it is crucial to employ a model that aligns with the expected budgets to facilitate superior synthetic performance. In this paper, we meticulously analyze the diffusion model and engineer a base model search approach, denoted "DiffNAS". Specifically, we leverage GPT-4 as a supernet to expedite the search, supplemented with a search memory to enhance the results. Moreover, we employ RFID as a proxy to promptly rank the experimental outcomes produced by GPT-4. We also adopt a rapid-convergence training strategy to boost search efficiency. Rigorous experimentation corroborates that our algorithm can augment the search efficiency by 2 times under GPT-based scenarios, while also attaining a performance of 2.82 with 0.37 improvement in FID on CIFAR10 relative to the benchmark IDDPM algorithm.

Human review

How Pith checks papers → · Full audit of this paper →

How the literature uses it

2 arXiv papers cite this in 3 sentences

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

Papers citing it · 1 sentence without a confident label

Cites it as background 2 papers

What it cites · 1 sentence without a confident label

Cites as background 2 papers
  • Currently, almost all mainstream diffusion models such as Guided Diffusion [ref] and Imagen [ref] leverage Unet as their foundation model to carry out denoising in the reverse process.

    Classifier-Free Diffusion Guidance · DiffNAS · 0.85

  • Diffusion models, which are extensively used in a variety of generative tasks including image generation [ref] , editing [ref] , and video generation [ref] , hinge on two processes: forward diffusion and reverse denoising.

    Imagen Video: High Definition Video Generation with Diffusion Models · Introduction · 0.96

Discussion

Know something about this paper? Leave a comment, or highlight any passage above to comment on it.