SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2311.07604 · ICLR · 2024

Finetuning Text-to-Image Diffusion Models for Fairness

Tianyu Pang, Mohan Kankanhalli, Min Lin, Yongkang Wong, Xudong Shen, Chao Du

arXiv · PDF · Open in the Atlas

Code that ran

2 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 2 failures of our instrument; 1 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2311.07604.json · badge: /api/ran/2311.07604.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
sail-sg/finetune-fair-diffusion canonical (classifier label)
classifier basis: classifier_2025
2 ran · 1 no recorded run
FunctionStatusWhere it lives
image_grid Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
sail-sg/finetune-fair-diffusion/exp-1-debias-gender/1-main-debias.py
code served (permissive licence) · get_code("c38c698ada5be90c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Image. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

from PIL import Image
import torch

torch.manual_seed(0)

# Create small test images with distinct sizes
# The function expects PIL Image objects with .size attribute and .paste method
img1 = Image.new('RGB', size=(64, 48), color=(255, 0, 0))
img2 = Image.new('RGB', size=(64, 48), color=(0, 255, 0))
img3 = Image.new('RGB', size=(64, 48), color=(0, 0, 255))
img4 = Image.new('RGB', size=(64, 48), color=(255, 255, 0))

imgs = [img1, img2, img3, img4]
rows = 2
cols = 2

ARGS = {
    'imgs': imgs,
    'rows': rows,
    'cols': cols
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
PIL, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Image — a Image output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Image output the recorded ndim 3 and dtype family int describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
make_grad_hook Ran
ran; our contract draft was wrong, not the code
sail-sg/finetune-fair-diffusion/exp-1-debias-gender/1-main-debias.py
code served (permissive licence) · get_code("51485529dc5e4917")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned function. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function make_grad_hook takes a single parameter 'coef'
# It returns a lambda function that multiplies its input by coef
# coef should be a numeric value (int, float, or tensor scalar)

ARGS = {
    "coef": 2.5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
customized_all_gather Not yet run
unverified (harvested, no recorded run)
sail-sg/finetune-fair-diffusion/exp-1-debias-gender/1-main-debias.py
code served (permissive licence) · get_code("a495b64cc3fbe810")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 3 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 3 later) and 0 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
image_grid 10 other papers
4 dated earlier, 6 later
2303.17870 (2023-03)
GlyphDraw: Seamlessly Rendering Text with Intricate Spatial Structures in Text-to-Image Generation
2310.04816 (2023-10)
Hacking Generative Models with Differentiable Network Bending
2310.06313 (2023-10)
Advancing Pose-Guided Image Synthesis with Progressive Conditional Diffusion Models
and 7 more not named here
customized_all_gather 1 other paper
1 later
Jung_Multi-Group_Proportional_Representations_for_Text-to-Image_Models_CVPR_2025_paper (2025 (year only))
Multi-Group Proportional Representation for Text-to-Image Models
make_grad_hook 1 other paper
1 later
Jung_Multi-Group_Proportional_Representations_for_Text-to-Image_Models_CVPR_2025_paper (2025 (year only))
Multi-Group Proportional Representation for Text-to-Image Models

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The rapid adoption of text-to-image diffusion models in society underscores an urgent need to address their biases. Without interventions, these biases could propagate a skewed worldview and restrict opportunities for minority groups. In this work, we frame fairness as a distributional alignment problem. Our solution consists of two main technical contributions: (1) a distributional alignment loss that steers specific characteristics of the generated images towards a user-defined target distribution, and (2) adjusted direct finetuning of diffusion model's sampling process (adjusted DFT), which leverages an adjusted gradient to directly optimize losses defined on the generated images. Empirically, our method markedly reduces gender, racial, and their intersectional biases for occupational prompts. Gender bias is significantly reduced even when finetuning just five soft tokens. Crucially, our method supports diverse perspectives of fairness beyond absolute equality, which is demonstrated by controlling age to a 75% young and 25% old distribution while simultaneously debiasing gender and race. Finally, our method is scalable: it can debias multiple concepts at once by simply including these prompts in the finetuning data. We share code and various fair diffusion model adaptors at https://sail-sg.github.io/finetune-fair-diffusion/.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2311.07604")
get_code_for_paper("2311.07604")
have("2311.07604")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2311.07604.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2311.07604.svg)](https://syntology.ai/paper/2311.07604)

Connect an agent — have() is free.