SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2405.18330 · NeurIPS · 2024

Frustratingly Easy Test-Time Adaptation of Vision-Language Models

Elisa Ricci, Massimiliano Mancini, Gianni Franchi, Matteo Farina, Giovanni Iacca

arXiv · PDF · Open in the Atlas

Code that ran

4 of 8 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; 4 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/2405.18330.json · badge: /api/ran/2405.18330.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
farinamatteo/zero canonical (classifier label)
classifier basis: census
4 ran · 4 no recorded run
FunctionStatusWhere it lives
basic_clean Ran
ran; our contract draft was wrong, not the code
farinamatteo/zero/clip/simple_tokenizer.py
pointer only (licence: MIT); the stored copy get_code serves, from ltpwy/MSCI, is pointer only (licence: NONE) · get_code("98f385d847636a3e")
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 ltpwy/MSCI, 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.

confidence_filter Ran
ran; our contract draft was wrong, not the code
farinamatteo/zero/ttas/base.py
code served (permissive licence) · get_code("23f0fc9cbefff039")
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 2, dtype family float, shape (1, 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.functional as F

torch.manual_seed(0)

# Create small tensors
batch_size = 2
num_classes = 4

# logits: raw model outputs, shape (batch, num_classes)
logits = torch.randn(batch_size, num_classes)

# probs: probability distribution, shape (batch, num_classes)
# Must be valid probabilities (sum to 1)
probs = F.softmax(logits, dim=1)

# top: float between 0 and 1, fraction of lowest-entropy samples to keep
top = 0.5

# return_idx: boolean flag
return_idx = False

ARGS = {
    "logits": logits,
    "probs": probs,
    "top": top,
    "return_idx": return_idx,
}
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 2, dtype family float, shape (1, 4) — shape (1, 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
get_pairs Ran
ran; our contract draft was wrong, not the code
farinamatteo/zero/clip/simple_tokenizer.py
code served (permissive licence) · get_code("d919ae32e5e4e616")
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 set. 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 get_pairs takes a word parameter
# A word is represented as a tuple of symbols (variable-length strings)
# The function iterates through the word and creates pairs of consecutive symbols

word = ("hello", "world", "test")

ARGS = {
    "word": word
}
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 set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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
whitespace_clean Ran
ran; our contract draft was wrong, not the code
farinamatteo/zero/clip/simple_tokenizer.py
pointer only (licence: MIT); the stored copy get_code serves, from ltpwy/MSCI, is pointer only (licence: NONE) · get_code("9542161e9640b858")
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 ltpwy/MSCI, 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.

build_model Not yet run
unverified (harvested, no recorded run)
farinamatteo/zero/maple_clip/model.py
code served (permissive licence) · get_code("48ac2bd4dcbfb717")
How it ran

no invocation recorded.

build_model Not yet run
unverified (harvested, no recorded run)
farinamatteo/zero/clip/model.py
pointer only (licence: MIT); the stored copy get_code serves, from GIT-LJc/LogitGap, is pointer only (licence: NONE) · get_code("aa56b568a90f8516")
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 GIT-LJc/LogitGap, 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.

load Not yet run
unverified (harvested, no recorded run)
farinamatteo/zero/maple_clip/clip.py
code served (permissive licence) · get_code("fbf8c0143d9c48e3")
How it ran

no invocation recorded.

load Not yet run
unverified (harvested, no recorded run)
farinamatteo/zero/clip/clip.py
pointer only (licence: MIT); the stored copy get_code serves, from GIT-LJc/LogitGap, is pointer only (licence: NONE) · get_code("88d0e8acec4aead0")
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 GIT-LJc/LogitGap, 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.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 7 are also present in other papers whose code we hold (7 with at least one paper dated earlier, 5 the same month, 7 later) and 1 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
get_pairs 258 other papers
143 dated earlier, 8 the same month, 99 later, 7 same year, order not determinable, 1 undated · 4 only as an assembly brick
1802.06474 (2018-02)
A Closed-form Solution to Photorealistic Image Stylization
1803.01128 (2018-03)
Seq2Sick: Evaluating the Robustness of Sequence-to-Sequence Models with Adversarial Examples
1809.00125 (2018-09)
Simple Fusion: Return of the Language Model
and 255 more not named here
whitespace_clean 231 other papers
124 dated earlier, 8 the same month, 91 later, 7 same year, order not determinable, 1 undated · 4 only as an assembly brick
1901.07441 (2019-01)
PadChest: A large chest x-ray image dataset with multi-label annotated reports
2107.06383 (2021-07)
How Much Can CLIP Benefit Vision-and-Language Tasks?
2108.10904 (2021-08)
SimVLM: Simple Visual Language Model Pretraining with Weak Supervision
and 228 more not named here
basic_clean 218 other papers
124 dated earlier, 8 the same month, 78 later, 7 same year, order not determinable, 1 undated · 4 only as an assembly brick
1901.07441 (2019-01)
PadChest: A large chest x-ray image dataset with multi-label annotated reports
2107.06383 (2021-07)
How Much Can CLIP Benefit Vision-and-Language Tasks?
2108.10904 (2021-08)
SimVLM: Simple Visual Language Model Pretraining with Weak Supervision
and 215 more not named here
load 88 other papers
44 dated earlier, 4 the same month, 33 later, 6 same year, order not determinable, 1 undated
2203.05557 (2022-03)
Conditional Prompt Learning for Vision-Language Models
2206.01986 (2022-06)
Delving into the Openness of CLIP
2207.12819 (2022-07)
S-Prompts Learning with Pre-trained Transformers: An Occam's Razor for Domain Incremental Learning
and 85 more not named here
build_model 30 other papers
12 dated earlier, 1 the same month, 14 later, 3 same year, order not determinable
2210.03117 (2022-10)
MaPLe: Multi-modal Prompt Learning
2212.03640 (2022-12)
Fine-tuned CLIP Models are Efficient Video Learners
2304.04704 (2023-04)
Prompt Pre-Training with Twenty-Thousand Classes for Open-Vocabulary Visual Recognition
and 27 more not named here
build_model 9 other papers
6 dated earlier, 3 later
2305.18010 (2023-05)
Test-Time Adaptation with CLIP Reward for Zero-Shot Generalization in Vision-Language Models
2311.15243 (2023-11)
ID-like Prompt Learning for Few-Shot Out-of-Distribution Detection
2401.15657 (2024-01)
Data-Free Generalized Zero-Shot Learning
and 6 more not named here
load 8 other papers
5 dated earlier, 3 later
2311.15243 (2023-11)
ID-like Prompt Learning for Few-Shot Out-of-Distribution Detection
2401.15657 (2024-01)
Data-Free Generalized Zero-Shot Learning
2403.12952 (2024-03)
Just Shift It: Test-Time Prototype Shifting for Zero-Shot Generalization with Vision-Language Models
and 5 more not named here

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

Vision-Language Models seamlessly discriminate among arbitrary semantic categories, yet they still suffer from poor generalization when presented with challenging examples. For this reason, Episodic Test-Time Adaptation (TTA) strategies have recently emerged as powerful techniques to adapt VLMs in the presence of a single unlabeled image. The recent literature on TTA is dominated by the paradigm of prompt tuning by Marginal Entropy Minimization, which, relying on online backpropagation, inevitably slows down inference while increasing memory. In this work, we theoretically investigate the properties of this approach and unveil that a surprisingly strong TTA method lies dormant and hidden within it. We term this approach ZERO (TTA with "zero" temperature), whose design is both incredibly effective and frustratingly simple: augment N times, predict, retain the most confident predictions, and marginalize after setting the Softmax temperature to zero. Remarkably, ZERO requires a single batched forward pass through the vision encoder only and no backward passes. We thoroughly evaluate our approach following the experimental protocol established in the literature and show that ZERO largely surpasses or compares favorably w.r.t. the state-of-the-art while being almost 10× faster and 13× more memory friendly than standard Test-Time Prompt Tuning. Thanks to its simplicity and comparatively negligible computation, ZERO can serve as a strong baseline for future work in this field. The code is available.

For agents

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

get_harvested_code_for_paper("2405.18330")
get_code_for_paper("2405.18330")
have("2405.18330")

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

curl https://syntology.ai/api/ran/2405.18330.json

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

[![Syntology run record](https://syntology.ai/api/ran/2405.18330.svg)](https://syntology.ai/paper/2405.18330)

Connect an agent — have() is free.