SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.19949 · CVPR · 2024

FairCLIP: Harnessing Fairness in Vision-Language Learning

Yu Tian, Yan Luo, Yi Fang, Min Shi, Ava Kouhana, Tobias Elze, Mengyu Wang, Hao Huang, Muhammad Afzal, Muhammad Osama Khan, Shuaihang Yuan, Luo Song

arXiv · PDF · Open in the Atlas

Code that ran

2 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 0 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2403.19949.json · badge: /api/ran/2403.19949.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
purdue-m2/robust_fairness_for_medical_image community (classifier label)
classifier basis not recorded on the link
1 ran · 2 no recorded run
Harvard-Ophthalmology-AI-Lab/FairCLIP marked official in Papers with Code
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
Adversary_Net Ran
ran on a synthesized input · constructed an object; no result was computed
Harvard-Ophthalmology-AI-Lab/FairCLIP/FairCLIP/src/modules.py
code served (permissive licence) · get_code("c77ee92e8218db6e")
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, through the batch API) and returned an instance of its own class, Adversary_Net: the run constructed the object and did not call it, so no result was computed. 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
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "n_sensitive": 5,
    "n_hidden": 32,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Adversary_Net — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Adversary_Net 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
truncate_note Ran
ran on a synthesized input
purdue-m2/robust_fairness_for_medical_image/mae/engine_finetune.py
code served (permissive licence) · get_code("3ead4b7caf1ee67e")
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, through the batch API) and returned str. 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

ARGS = {
    "note": "This is a very long note that contains many words and will definitely exceed the maximum length allowed by the truncate_note function so it will need to be truncated at a word boundary rather than in the middle of a word to maintain readability and proper formatting for the output result.",
    "max_length": 170
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
compute_gradcam Not yet run
unverified (harvested, no recorded run)
purdue-m2/robust_fairness_for_medical_image/LAVIS/lavis/models/blip_models/blip_image_text_matching.py
pointer only (licence: MIT); the stored copy get_code serves, from jackwaky/QuRe, is pointer only (licence: NONE) · get_code("16b8bc0c4c1c1c87")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from jackwaky/QuRe, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

weight_estimation Not yet run
unverified (harvested, no recorded run)
purdue-m2/robust_fairness_for_medical_image/RobustFairCLIP/finetune_RobustFairCLIP.py
code served (permissive licence) · get_code("84d9d90b28891b49")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 3 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
compute_gradcam 2 other papers
2 later
2406.18113 (2024-06)
Chrono: A Simple Blueprint for Representing Time in MLLMs
2507.12416 (2025-07)
QuRe: Query-Relevant Retrieval through Hard Negative Sampling in Composed Image Retrieval

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

Fairness is a critical concern in deep learning, especially in healthcare, where these models influence diagnoses and treatment decisions. Although fairness has been investigated in the vision-only domain, the fairness of medical vision-language (VL) models remains unexplored due to the scarcity of medical VL datasets for studying fairness. To bridge this research gap, we introduce the first fair visionlanguage medical dataset (Harvard-FairVLMed) that provides detailed demographic attributes, ground-truth labels, and clinical notes to facilitate an in-depth examination of fairness within VL foundation models. Using Harvard-FairVLMed, we conduct a comprehensive fairness analysis of two widely-used VL models (CLIP and BLIP2), pretrained on both natural and medical domains, across four different protected attributes. Our results highlight significant biases in all VL models, with Asian, Male, Non-Hispanic, and Spanish being the preferred subgroups across the protected attributes of race, gender, ethnicity, and language, respectively. In order to alleviate these biases, we propose FairCLIP, an optimal-transport-based approach that achieves a favorable trade-off between performance and fairness by reducing the Sinkhorn distance between the overall sample distribution and the distributions corresponding to each demographic group. As the first VL dataset of its kind, Harvard-FairVLMed holds the potential to catalyze advancements in the development of machine learning models that are both ethically aware and clinically effective. Our dataset and code are available at https://ophai.hms.harvard.edu/datasets/ harvard-fairvlmed10k.

For agents

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

get_harvested_code_for_paper("2403.19949")
get_code_for_paper("2403.19949")
have("2403.19949")

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

curl https://syntology.ai/api/ran/2403.19949.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.19949.svg)](https://syntology.ai/paper/2403.19949)

Connect an agent — have() is free.