Papers › TTD: Text-Tag Self-Distillation Enhancing Image-Text Alignment in CLIP to Alleviate...

TTD: Text-Tag Self-Distillation Enhancing Image-Text Alignment in CLIP to Alleviate Single Tag Bias

30 Mar 2024arXiv:2404.00384archive 2025-07-28

Sanghyun Jo, Soohyun Ryu, Sungyub Kim, Eunho Yang, KyungSu Kim

We identify a critical bias in contemporary CLIP-based models, which we denote as single tag bias. This bias manifests as a disproportionate focus on a singular tag (word) while neglecting other pertinent tags, stemming from CLIP's text embeddings that prioritize one specific tag in image-text relationships. When deconstructing text into individual tags, only one tag tends to have high relevancy with CLIP's image embedding, leading to biased tag relevancy. In this paper, we introduce a novel two-step fine-tuning approach, Text-Tag Self-Distillation (TTD), to address this challenge. TTD first extracts image-relevant tags from text based on their similarity to the nearest pixels then employs a self-distillation strategy to align combined masks with the text-derived mask. This approach ensures the unbiased image-text alignment of the CLIP-based models using only image-text pairs without necessitating additional supervision. Our technique demonstrates model-agnostic improvements in multi-tag classification and segmentation tasks, surpassing competing methods that rely on external resources. The code is available at https://github.com/shjo-april/TTD.

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 1 code sample harvested from 1 repository linked to this paper: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; failures of Syntology's instrument, not of the code. Every run on this page is a failure of Syntology's instrument, not of the code.

By repository: official repository: 1 sample from 1 repository, 1 ran (0 with no instrument failure; 1 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

shjo-april/TTD officialmentioned in papermentioned on GitHubSyntology: 1 sample, 1 ran (1 instrument-failed) · 0 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

1 sample harvested: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Every run on this page is a failure of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · our draft was wrong

Licence: 1 of the 1 sample is pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from shjo-april/TTD. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

shjo-april/TTD official repository 1 sample, 1 ran (1 instrument-failed) · 0 unverified
load in 392 papers core/tcl/clip/clip.py ran · our draft was wrong no licence file found · pointer only · 6901388a150b1941 · report fixture: returned MockCLIPModel · level recorded 2026-09-22

Tasks

Multi-Label Text ClassificationOpen Vocabulary Semantic SegmentationSemantic SegmentationTAGUnsupervised Semantic Segmentation with Language-image Pre-training

Datasets

Introduced by this paper, per the archive.

CC3M-TagMask

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Multi-Label Text Classification CC3M-TagMask TTD (w/ fine-tuning) Accuracy 88.6 #1 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/ fine-tuning) F1 82.8 #1 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/ fine-tuning) Precision 88.3 #1 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/ fine-tuning) Recall 78.0 #1 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/ fine-tuning) mAP 93.7 #1 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/o fine-tuning) Accuracy 91.0 #2 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/o fine-tuning) F1 78.5 #2 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/o fine-tuning) Precision 82.9 #2 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/o fine-tuning) Recall 74.5 #2 of 6 Archive leaderboard report
Multi-Label Text Classification CC3M-TagMask TTD (w/o fine-tuning) mAP 90.3 #2 of 6 Archive leaderboard report
Open Vocabulary Semantic Segmentation ADE20K-150 TTD (TCL) mIoU 17.0 #21 of 23 Archive leaderboard report
Open Vocabulary Semantic Segmentation ADE20K-150 TTD (MaskCLIP) mIoU 12.7 #23 of 23 Archive leaderboard report
Open Vocabulary Semantic Segmentation COCO-Stuff-171 TTD (TCL) mIoU 23.7 #4 of 7 Archive leaderboard report
Open Vocabulary Semantic Segmentation COCO-Stuff-171 TTD (MaskCLIP) mIoU 19.4 #7 of 7 Archive leaderboard report
Open Vocabulary Semantic Segmentation Cityscapes TTD (TCL) mIoU 32.0 #3 of 5 Archive leaderboard report
Open Vocabulary Semantic Segmentation Cityscapes TTD (MaskCLIP) mIoU 27.0 #5 of 5 Archive leaderboard report
Open Vocabulary Semantic Segmentation PASCAL Context-59 TTD (TCL) mIoU 37.4 #20 of 24 Archive leaderboard report
Open Vocabulary Semantic Segmentation PASCAL Context-59 TTD (MaskCLIP) mIoU 31.0 #23 of 24 Archive leaderboard report
Semantic Segmentation CC3M-TagMask TTD (TCL) mIoU 65.5 #1 of 4 Archive leaderboard report
Semantic Segmentation CC3M-TagMask TTD (MaskCLIP) mIoU 50.2 #3 of 4 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training ADE20K TTD (TCL) Mean IoU (val) 17.0 #8 of 13 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training ADE20K TTD (MaskCLIP) Mean IoU (val) 12.7 #10 of 13 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training COCO-Object TTD (TCL) mIoU 37.4 #4 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training COCO-Object TTD (MaskCLIP) mIoU 26.5 #10 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training COCO-Stuff-171 TTD (TCL) mIoU 23.7 #6 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training COCO-Stuff-171 TTD (MaskCLIP) mIoU 19.4 #9 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training Cityscapes val TTD (MaskCLIP) mIoU 32.0 #5 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training Cityscapes val TTD (TCL) mIoU 27.0 #7 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training PASCAL Context-59 TTD (TCL) mIoU 37.4 #6 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training PASCAL Context-59 TTD (MaskCLIP) mIoU 31.0 #9 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training PASCAL VOC TTD (TCL) mIoU 61.1 #6 of 10 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training PASCAL VOC TTD (MaskCLIP) mIoU 43.1 #9 of 10 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

ALIGNFocus

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections