SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.12128 · ECCV · 2024

Distribution Alignment for Fully Test-Time Adaptation with Dynamic Online Data Streams

Konstantinos Plataniotis, Yang Wang, Yan Wu, Zhixiang Chi, Z Liu, Li Gu, Ziqiang Wang

arXiv · PDF · Open in the Atlas

Code that ran

13 of 17 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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 12 failures of our instrument; 4 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/2407.12128.json · badge: /api/ran/2407.12128.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
WZq975/DA-TTA canonical (classifier label)
classifier basis: census
13 ran · 4 no recorded run
FunctionStatusWhere it lives
complete_data_dir_path Ran
ran; our contract draft was wrong, not the code
WZq975/DA-TTA/conf.py
code served (permissive licence) · get_code("00900164f55d3e7c")
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 os

ARGS = {
    'root': '/tmp/datasets',
    'dataset_name': 'imagenet'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — 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
create_file_list Ran
ran; our contract draft was wrong, not the code
WZq975/DA-TTA/datasets/imagenet_subsets.py
code served (permissive licence) · get_code("ae33aa95d68dffd6")
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 list, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 os
import tempfile
import shutil

# Create a temporary directory structure that mimics ImageNet
temp_dir = tempfile.mkdtemp()
imagenet_root = temp_dir

# Create the split directories and class folders with fake JPEG files
split = 'train'
split_dir = os.path.join(imagenet_root, split)
os.makedirs(split_dir, exist_ok=True)

# Create a few class directories with fake JPEG files
class_dirs = ['n01440764', 'n01443537', 'n01484850']
for class_dir in class_dirs:
    class_path = os.path.join(split_dir, class_dir)
    os.makedirs(class_path, exist_ok=True)
    # Create a few fake JPEG files in each class directory
    for i in range(2):
        fake_jpeg_path = os.path.join(class_path, f'image_{i}.JPEG')
        with open(fake_jpeg_path, 'w') as f:
            f.write('fake jpeg data')

ARGS = {
    'imagenet_root': imagenet_root,
    'imagenet_class_dirs': class_dirs,
    'split': split
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
entropy_energy Ran
ran; our contract draft was wrong, not the code
WZq975/DA-TTA/methods/lame.py
code served (permissive licence) · get_code("473a24407468b75e")
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 Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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)

# Y: tensor of probabilities/activations
Y = torch.rand(2, 4, requires_grad=True)

# unary: unary potentials/energies
unary = torch.randn(2, 4)

# pairwise: pairwise potentials/energies
pairwise = torch.randn(2, 4)

# bound_lambda: scalar weight parameter
bound_lambda = 0.5

ARGS = {
    "Y": Y,
    "unary": unary,
    "pairwise": pairwise,
    "bound_lambda": bound_lambda
}
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 Tensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
get_num_classes Ran
ran, honoured the contract we drafted
WZq975/DA-TTA/conf.py
code served (permissive licence) · get_code("ad7ee34b3774e46a")
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 int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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 = {
    'dataset_name': 'cifar10'
}
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 int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
marginal_entropy Ran
ran; our contract draft was wrong, not the code
WZq975/DA-TTA/methods/memo.py
pointer only (licence: MIT); the stored copy get_code serves, from dmn-sjk/intransigent_teacher, is pointer only (licence: NONE) · get_code("8a657c2641647a46")
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 dmn-sjk/intransigent_teacher, 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. A run is recorded.

softmax_entropy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WZq975/DA-TTA/methods/datta.py
pointer only (licence: MIT); the stored copy get_code serves, from Feliciaxyao/ICML2024-FSTTA, is pointer only (licence: NONE) · get_code("3078648c75e3e1e3")
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 Feliciaxyao/ICML2024-FSTTA, 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. A run is recorded.

softmax_entropy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WZq975/DA-TTA/methods/eata.py
code served (permissive licence) · get_code("7b9d88d29b50aca6")
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 Tensor, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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)

# Create a small batch of logits with shape (batch_size, num_classes)
# batch_size = 2, num_classes = 4
x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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 Tensor, ndim 1, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
softmax_entropy_cifar Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WZq975/DA-TTA/methods/cotta.py
code served (permissive licence) · get_code("f0cf18b3022925f3")
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 Tensor, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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)

# Create small tensors for x and x_ema
# Both need to support .softmax(1), .log_softmax(1), and .sum(1)
# These are standard tensor methods, so any tensor works
# Using shape (2, 4) for batch_size=2, num_classes=4

x = torch.randn(2, 4, requires_grad=True)
x_ema = torch.randn(2, 4)

ARGS = {
    "x": x,
    "x_ema": x_ema,
}
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 Tensor, ndim 1, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
softmax_entropy_imagenet Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WZq975/DA-TTA/methods/cotta.py
code served (permissive licence) · get_code("e2785d59d7e26ba2")
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 Tensor, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 takes two parameters: x and x_ema
# Both are used with:
# - .softmax(1) method
# - .log_softmax(1) method
# - element-wise multiplication
# - .sum(1) method
# These are all standard tensor operations, so we need torch.Tensor objects

# Create two small tensors with batch size 2 and 4 logits (for ImageNet-like setup)
x = torch.randn(2, 4)
x_ema = torch.randn(2, 4)

ARGS = {
    "x": x,
    "x_ema": x_ema,
}
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 Tensor, ndim 1, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
stat_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WZq975/DA-TTA/methods/datta.py
code served (permissive licence) · get_code("155e6f06ad63c237")
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 Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type, ndim and shape 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)

# The function expects:
# 1. stat: a list of tuples, where each tuple contains (mean_tensor, var_tensor)
# 2. source_stat: a dict with 'means' and 'vars' keys containing lists of tensors

# Build stat - list of tuples with mean and var tensors
# Each element is a tuple of (mean_tensor, var_tensor)
# Batch size is 2, feature dimension is 4
stat = [
    (torch.randn(2, 4), torch.randn(2, 4)),  # First stat pair
    (torch.randn(2, 4), torch.randn(2, 4)),  # Second stat pair
    (torch.randn(2, 4), torch.randn(2, 4)),  # Third stat pair
]

# Build source_stat - dict with means and vars
# Each mean/var should be a 1D tensor that will be unsqueezed and repeated
source_stat = {
    'means': [
        torch.randn(4),  # Shape for first mean
        torch.randn(4),  # Shape for second mean
        torch.randn(4),  # Shape for third mean
    ],
    'vars': [
        torch.randn(4),  # Shape for first var
        torch.randn(4),  # Shape for second var
        torch.randn(4),  # Shape for third var
    ]
}

ARGS = {
    'stat': stat,
    'source_stat': source_stat,
}
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 Tensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
tta Ran
ran; our contract draft was wrong, not the code
WZq975/DA-TTA/methods/memo.py
code served (permissive licence) · get_code("eaa0b5d0d74c3ab3")
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 Tensor, ndim 4, dtype family float, shape (2, 3, 4, 4) as a later run recorded it. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type, ndim and shape 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 numpy as np
from torchvision import transforms
from PIL import Image

torch.manual_seed(0)

# Create a small image tensor (1, 3, 4, 4) - batch of 1, 3 channels, 4x4 spatial
image_tensor = torch.rand(1, 3, 4, 4)

# Number of augmentations
n_augmentations = 2

# Create a simple augmentation function that takes a PIL Image and returns a tensor
class SimpleAugmentation:
    def __call__(self, pil_image):
        # Convert PIL image to tensor
        return transforms.ToTensor()(pil_image)

aug = SimpleAugmentation()

ARGS = {
    "image": image_tensor,
    "n_augmentations": n_augmentations,
    "aug": aug,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (2, 3, 4, 4) — shape (2, 3, 4, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
update_ema_variables Ran
ran; our contract draft was wrong, not the code
WZq975/DA-TTA/methods/cotta.py
pointer only (licence: MIT); the stored copy get_code serves, from joe817/gcal, is pointer only (licence: NONE) · get_code("a77d556c0aea19c1")
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 joe817/gcal, 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. A run is recorded.

update_model_probs Ran
ran; our contract draft was wrong, not the code
WZq975/DA-TTA/methods/eata.py
code served (permissive licence) · get_code("182d3657a9eebb0d")
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 Tensor, ndim 1, dtype family float, shape (4,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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)

# Case 1: current_model_probs is None, new_probs is non-empty
# This will return new_probs.mean(0)
current_model_probs_1 = None
new_probs_1 = torch.randn(2, 4)

# Case 2: current_model_probs exists, new_probs is empty
# This will return current_model_probs
current_model_probs_2 = torch.randn(4)
new_probs_2 = torch.randn(0, 4)

# Case 3: both exist, update with weighted average
current_model_probs_3 = torch.randn(4)
new_probs_3 = torch.randn(2, 4)

# We'll use case 3 as the main test case
ARGS = {
    "current_model_probs": torch.randn(4),
    "new_probs": torch.randn(2, 4)
}
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 Tensor, ndim 1, dtype family float, shape (4,) — shape (4,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
batch_norm Not yet run
unverified (harvested, no recorded run)
WZq975/DA-TTA/methods/datta.py
code served (permissive licence) · get_code("fae69c4390cf3b2c")
How it ran

no invocation recorded.

create_imagenet_subset Not yet run
unverified (harvested, no recorded run)
WZq975/DA-TTA/datasets/imagenet_subsets.py
code served (permissive licence) · get_code("04cb16a031d4e8b2")
How it ran

no invocation recorded.

get_domain_sequence Not yet run
unverified (harvested, no recorded run)
WZq975/DA-TTA/conf.py
code served (permissive licence) · get_code("cfc577ab97377255")
How it ran

no invocation recorded.

laplacian_optimization Not yet run
unverified (harvested, no recorded run)
WZq975/DA-TTA/methods/lame.py
code served (permissive licence) · get_code("d2597d43cbd2fbca")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 17 distinct functions with an identity, 12 are also present in other papers whose code we hold (12 with at least one paper dated earlier, 1 the same month, 7 later) and 5 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
softmax_entropy 22 other papers
14 dated earlier, 1 the same month, 6 later, 1 undated · 4 only as an assembly brick
2006.10726 (2020-06)
TENT: FULLY TEST-TIME ADAPTATION BY ENTROPY MINIMIZATION
2105.08714 (2021-05)
Fighting Gradients with Gradients: Dynamic Defenses against Adversarial Attacks
2110.14038 (2021-10)
Robustness of Graph Neural Networks at Scale
and 19 more not named here
update_ema_variables 14 other papers
10 dated earlier, 4 later · 2 only as an assembly brick
2203.13591 (2022-03)
Continual Test-Time Domain Adaptation
2212.09713 (2022-12)
A Probabilistic Framework for Lifelong Test-Time Adaptation
2306.04344 (2023-06)
ViDA: Homeostatic Visual Domain Adapter for Continual Test Time Adaptation
and 11 more not named here
softmax_entropy 8 other papers
6 dated earlier, 1 later, 1 undated
2006.10726 (2020-06)
TENT: FULLY TEST-TIME ADAPTATION BY ENTROPY MINIMIZATION
2204.02610 (2022-04)
Efficient Test-Time Model Adaptation without Forgetting
2311.14402 (2023-11)
TEA: Test-time Energy Adaptation
and 5 more not named here
update_model_probs 8 other papers
4 dated earlier, 3 later, 1 undated
2204.02610 (2022-04)
Efficient Test-Time Model Adaptation without Forgetting
2309.14949 (2023-09)
Towards Real-World Test-Time Adaptation: Tri-Net Self-Training with Balanced Normalization
2311.14402 (2023-11)
TEA: Test-time Energy Adaptation
and 5 more not named here
entropy_energy 5 other papers
4 dated earlier, 1 later
2308.10297 (2023-08)
DomainAdaptor: A Novel Approach to Test-time Adaptation
2309.14949 (2023-09)
Towards Real-World Test-Time Adaptation: Tri-Net Self-Training with Balanced Normalization
2401.08328 (2024-01)
Un-Mixing Test-Time Normalization Statistics: Combatting Label Temporal Correlation
and 2 more not named here
create_file_list 3 other papers
2 dated earlier, 1 same year, order not determinable
2310.20199 (2023-10)
In Search of Lost Online Test-time Adaptation: A Survey
2312.09486 (2023-12)
Unraveling Batch Normalization for Realistic Test-Time Adaptation
aaai_29436 (2024 (year only))
Unraveling Batch Normalization for Realistic Test-Time Adaptation
get_domain_sequence 3 other papers
2 dated earlier, 1 same year, order not determinable
2310.20199 (2023-10)
In Search of Lost Online Test-time Adaptation: A Survey
2312.09486 (2023-12)
Unraveling Batch Normalization for Realistic Test-Time Adaptation
aaai_29436 (2024 (year only))
Unraveling Batch Normalization for Realistic Test-Time Adaptation
laplacian_optimization 3 other papers
2 dated earlier, 1 later
2401.08328 (2024-01)
Un-Mixing Test-Time Normalization Statistics: Combatting Label Temporal Correlation
2405.00754 (2024-05)
CLIPArTT: Adaptation of CLIP to New Domains at Test Time
2601.16240 (2026-01)
Test-Time Adaptation for Speech Emotion Recognition
marginal_entropy 3 other papers
2 dated earlier, 1 later
2310.18816 (2023-10)
Adaptive Test-Time Personalization for Federated Learning
2404.01351 (2024-04)
AETTA: Label-Free Accuracy Estimation for Test-Time Adaptation
2609.02507 (2026-09)
Rethinking the Teacher-Student Framework for Test-Time Adaptation
get_num_classes 2 other papers
1 dated earlier, 1 same year, order not determinable
2312.09486 (2023-12)
Unraveling Batch Normalization for Realistic Test-Time Adaptation
aaai_29436 (2024 (year only))
Unraveling Batch Normalization for Realistic Test-Time Adaptation
softmax_entropy_cifar 2 other papers
2 dated earlier
2203.13591 (2022-03)
Continual Test-Time Domain Adaptation
2310.20199 (2023-10)
In Search of Lost Online Test-time Adaptation: A Survey
softmax_entropy_imagenet 2 other papers
2 dated earlier
2203.13591 (2022-03)
Continual Test-Time Domain Adaptation
2310.20199 (2023-10)
In Search of Lost Online Test-time Adaptation: A Survey

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

Given a model trained on source data, Test-Time Adaptation (TTA) enables adaptation and inference in test data streams with domain shifts from the source. Current methods predominantly optimize the model for each incoming test data batch using self-training loss. While these methods yield commendable results in ideal test data streams, where batches are independently and identically sampled from the target distribution, they falter under more practical test data streams that are not independent and identically distributed (non-i.i.d.). The data batches in a non-i.i.d. stream display prominent label shifts relative to each other. It leads to conflicting optimization objectives among batches during the TTA process. Given the inherent risks of adapting the source model to unpredictable test-time distributions, we reverse the adaptation process and propose a novel Distribution Alignment loss for TTA. This loss guides the distributions of test-time features back towards the source distributions, which ensures compatibility with the well-trained source model and eliminates the pitfalls associated with conflicting optimization objectives. Moreover, we devise a domain shift detection mechanism to extend the success of our proposed TTA method in the continual domain shift scenarios. Our extensive experiments validate the logic and efficacy of our method. On six benchmark datasets, we surpass existing methods in non-i.i.d. scenarios and maintain competitive performance under the ideal i.i.d. assumption.

For agents

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

get_harvested_code_for_paper("2407.12128")
get_code_for_paper("2407.12128")
have("2407.12128")

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

curl https://syntology.ai/api/ran/2407.12128.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.12128.svg)](https://syntology.ai/paper/2407.12128)

Connect an agent — have() is free.