Papers › The Golden Subspace: Where Efficiency Meets Generalization in Continual Test-Time Adaptation

The Golden Subspace: Where Efficiency Meets Generalization in Continual Test-Time Adaptation

23 Mar 2026arXiv:2603.21928added by Syntology

Guannan Lai, Da-Wei Zhou, Zhenguo Li, Han-Jia Ye

Title, abstract, authors and date from arXiv's metadata (CC0); this paper is not in the Papers with Code archive (frozen 2025-07-28).

Continual Test-Time Adaptation (CTTA) aims to enable models to adapt online to unlabeled data streams under distribution shift without accessing source data. Existing CTTA methods face an efficiency-generalization trade-off: updating more parameters improves adaptation but severely reduces online inference efficiency. An ideal solution is to achieve comparable adaptation with minimal feature updates; we call this minimal subspace the golden subspace. We prove its existence in a single-step adaptation setting and show that it coincides with the row space of the pretrained classifier. To enable online maintenance of this subspace, we introduce the sample-wise Average Gradient Outer Product (AGOP) as an efficient proxy for estimating the classifier weights without retraining. Building on these insights, we propose Guided Online Low-rank Directional adaptation (GOLD), which uses a lightweight adapter to project features onto the golden subspace and learns a compact scaling vector while the subspace is dynamically updated via AGOP. Extensive experiments on classification and segmentation benchmarks, including autonomous-driving scenarios, demonstrate that GOLD attains superior efficiency, stability, and overall performance. Our code is available at https://github.com/AIGNLAI/GOLD.

PaperPDFCode 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 13 code samples harvested from 1 repository linked to this paper: 12 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 2 honoured, 0 violated, 3 with no contract checked; 7 where Syntology's instrument failed) · 1 unverified. Where Syntology's instrument failed: 5 where our contract draft was wrong; 2 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: found in paper text by Syntology: 13 samples from 1 repository, 12 ran (5 with no instrument failure; 7 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 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

AIGNLAI/GOLD found in paper text by SyntologySyntology: 13 samples, 12 ran (7 instrument-failed) · 1 unverified report

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

13 samples harvested: 12 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 2 honoured, 0 violated, 3 with no contract checked; 7 where Syntology's instrument failed) · 1 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

2ran · honoured contract
5ran · our draft was wrong
2ran; the contract check could not run
3ran
1unverified

Licence: 13 of the 13 samples are 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 AIGNLAI/GOLD. “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. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

AIGNLAI/GOLD found in paper text by Syntology 13 samples, 12 ran (7 instrument-failed) · 1 unverified
complete_data_dir_path in 9 papers classification/conf.py ran · our draft was wrong no licence file found · pointer only · 53b1e10b7cc3d6de · report fixture: signature data_root_dir:str/0/str, dataset_name:str/0/str · returned str · level recorded 2026-09-25
create_file_list in 10 papers classification/datasets/imagenet_subsets.py ran · our draft was wrong no licence file found · pointer only · 0c4495cb2187be84 · report fixture: signature imagenet_root:str/0/str, imagenet_class_dirs:list/1/str, split:str/0/str · returned list · level recorded 2026-09-25
ds_name2pytorch_ds_name classification/conf.py ran · our draft was wrong no licence file found · pointer only · 22a042db3ce6f8bb · report fixture: signature ds_name:str/0/str · returned str · level recorded 2026-09-25
forward_decorator classification/methods/base.py ran · our draft was wrong no licence file found · pointer only · ef670fd2920b7ce2 · report fixture: signature fn:callable/0/? · returned function · level recorded 2026-09-25
get_num_classes in 26 papers classification/conf.py ran · honoured contract no licence file found · pointer only · 4d83f34a9766b4a9 · report fixture: signature dataset_name:str/0/str · returned int · level recorded 2026-09-25
kernel in 22 papers classification/methods/cmf.py ran no licence file found · pointer only · b086cfc2dfe368f4 · report fixture: signature model:callable/0/?, src_model:callable/0/?, bias:float/0/float, normalization_constant:float/0/float · returned Tensor · level recorded 2026-09-25
refine_predictions in 7 papers classification/methods/adacontrast.py ran no licence file found · pointer only · 6c5a8d4a69a975c2 · report fixture: signature features:Tensor/2/float, probs:Tensor/2/float, banks:dict/0/?, refine_method:str/0/str, dist_type:str/0/str, num_neighbors:int/0/int, gt_labels:Tensor/1/int · returned tuple · level recorded 2026-09-25
setup_adacontrast_optimizer classification/methods/adacontrast.py ran · our draft was wrong no licence file found · pointer only · ba822e549de6a087 · report fixture: signature model:callable/0/?, cfg:Config/0/? · returned SGD · level recorded 2026-09-25
soft_k_nearest_neighbors in 6 papers classification/methods/adacontrast.py ran no licence file found · pointer only · 110b394c4d611e0b · report fixture: signature features:Tensor/2/float, features_bank:Tensor/2/float, probs_bank:Tensor/2/float, dist_type:str/0/str, num_neighbors:int/0/int · returned tuple · level recorded 2026-09-25
softmax_entropy_cifar in 7 papers classification/methods/cotta.py ran; the contract check could not run (RAISES) no licence file found · pointer only · 991d0f3401682804 · report fixture: returned Tensor · level recorded 2026-09-25
softmax_entropy_imagenet in 7 papers classification/methods/cotta.py ran; the contract check could not run (RAISES) no licence file found · pointer only · 6bdfed740b3da7ce · report fixture: returned Tensor · level recorded 2026-09-25
update_model_probs in 12 papers
census: 4 implementations from 5 papers, 3 distinct outputs across 3 buckets, one shared input per bucket
classification/methods/cmf.py ran · honoured contract fingerprinted no licence file found · pointer only · 073ab5af80ddbe36 · report fixture: signature x_ema:Tensor/2/float, x:Tensor/2/float, momentum:float/0/float · returned Tensor · level recorded 2026-09-25
create_imagenet_subset in 8 papers classification/datasets/imagenet_subsets.py unverified no licence file found · pointer only · b723ea0fcdc2c948 · report

Results from the paper

The Papers with Code archive ends with its 2025-07-28 snapshot. This paper's arXiv identifier, 2603.21928, was issued in March 2026, after that date, so the archive has no leaderboard rows for it.

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