SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2210.00788 · 2022

Towards a Unified View on Visual Parameter-Efficient Transfer Learning

arXiv · PDF · Open in the Atlas

Code that ran

13 of 13 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; 2 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 9 failures of our instrument; 0 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/2210.00788.json · badge: /api/ran/2210.00788.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
bruceyo/V-PETL canonical (classifier label)
classifier basis not recorded on the link
13 ran · 0 no recorded run
FunctionStatusWhere it lives
build_dataset Ran
ran; our contract draft was wrong, not the code
bruceyo/V-PETL/util/datasets.py
pointer only (licence: MIT); the stored copy get_code serves, from facebookresearch/mae, is pointer only (licence: NOASSERTION) · get_code("27f8c4c706fd43ec")
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 facebookresearch/mae, and GitHub could not classify the licence of that repository (NOASSERTION). 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_transform Ran
ran; our contract draft was wrong, not the code
bruceyo/V-PETL/util/datasets.py
pointer only (licence: MIT); the stored copy get_code serves, from facebookresearch/mae, is pointer only (licence: NOASSERTION) · get_code("800cebf161a8aa26")
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 facebookresearch/mae, and GitHub could not classify the licence of that repository (NOASSERTION). 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.

convert_img Ran
ran on a synthesized input
bruceyo/V-PETL/datasets/volume_transforms.py
code served (permissive licence) · get_code("d0baef849d9ee588")
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 ndarray, ndim 3, dtype family int, shape (3, 5, 7) 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 numpy as np

# Test case 1: 3D array (H, W, C) - most common case
img_3d = np.random.randint(0, 256, size=(5, 7, 3), dtype=np.uint8)

# Test case 2: 2D array (H, W) - grayscale image
img_2d = np.random.randint(0, 256, size=(5, 7), dtype=np.uint8)

# We'll use the 3D case as the primary test
ARGS = {
    "img": img_3d
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (3, 5, 7) — shape (3, 5, 7) 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
create_logger Ran
ran; our contract draft was wrong, not the code
bruceyo/V-PETL/models/logger.py
pointer only (licence: MIT); the stored copy get_code serves, from ghrua/seqpe, is pointer only (licence: NONE) · get_code("80450600f7f09b0a")
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 ghrua/seqpe, 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.

crop_boxes Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bruceyo/V-PETL/datasets/video_transforms.py
pointer only (licence: MIT); the stored copy get_code serves, from CAMMA-public/SurgVLP, is pointer only (licence: NONE) · get_code("157585dc6ae8a498")
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 CAMMA-public/SurgVLP, 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.

crop_clip Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bruceyo/V-PETL/datasets/functional.py
code served (permissive licence) · get_code("bc39b8ed5fefa7eb")
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 4, dtype family int; 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 numpy as np
import PIL
from PIL import Image

# Create a small numpy array-based clip
np_clip = [
    np.random.randint(0, 256, (100, 120, 3), dtype=np.uint8),
    np.random.randint(0, 256, (100, 120, 3), dtype=np.uint8),
]

# Create a small PIL Image-based clip
pil_clip = [
    Image.new('RGB', (120, 100), color=(73, 109, 137)),
    Image.new('RGB', (120, 100), color=(200, 150, 100)),
]

# Test with numpy arrays
ARGS = {
    'clip': np_clip,
    'min_h': 10,
    'min_w': 15,
    'h': 50,
    'w': 60,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 4, dtype family int — 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
get_resize_sizes Ran
ran, honoured the contract we drafted
bruceyo/V-PETL/datasets/functional.py
code served (permissive licence) · get_code("2b53e457d5632b4e")
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 tuple, ndim 1, dtype family int; 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
import numpy as np

# Analyze the function: get_resize_sizes takes im_h (image height), 
# im_w (image width), and size (a target dimension). It uses basic 
# arithmetic operations on these numeric values and returns a tuple.

torch.manual_seed(0)

# Create simple numeric arguments
im_h = 480  # image height - distinct value
im_w = 640  # image width - distinct value
size = 256  # target resize dimension - distinct value

ARGS = {
    "im_h": im_h,
    "im_w": im_w,
    "size": size,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple 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
random_crop Ran
ran on a synthesized input
bruceyo/V-PETL/datasets/video_transforms.py
code served (permissive licence) · get_code("967f6f1f3e4b6d6d")
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 tuple. 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 numpy as np

torch.manual_seed(0)

# Create a small tensor for images: num_frames x channel x height x width
# Using distinct dimensions: 2 frames, 3 channels, 8 height, 6 width
images = torch.randn(2, 3, 8, 6)

# Create boxes: num_boxes x 4 (x1, y1, x2, y2 format)
# 2 boxes with coordinates within the image bounds
boxes = np.array([[1.0, 1.0, 4.0, 5.0], [2.0, 2.0, 5.0, 6.0]], dtype=np.float32)

# Crop size should be smaller than image dimensions
size = 4

ARGS = {
    "images": images,
    "size": size,
    "boxes": boxes,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is 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
random_short_side_scale_jitter Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bruceyo/V-PETL/datasets/video_transforms.py
pointer only (licence: MIT); the stored copy get_code serves, from CAMMA-public/SurgVLP, is pointer only (licence: NONE) · get_code("8b52c4f275af3fc5")
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 CAMMA-public/SurgVLP, 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.

resize_clip Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bruceyo/V-PETL/datasets/functional.py
code served (permissive licence) · get_code("981936c73d3cdd12")
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 4, dtype family int; 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 numbers
import cv2
import numpy as np
import PIL
from PIL import Image

# Create a small numpy array clip (list of frames)
frame1 = np.random.randint(0, 256, (480, 640, 3), dtype=np.uint8)
frame2 = np.random.randint(0, 256, (480, 640, 3), dtype=np.uint8)
np_clip = [frame1, frame2]

# Create a PIL Image clip
pil_frame1 = Image.fromarray(np.random.randint(0, 256, (480, 640, 3), dtype=np.uint8))
pil_frame2 = Image.fromarray(np.random.randint(0, 256, (480, 640, 3), dtype=np.uint8))
pil_clip = [pil_frame1, pil_frame2]

ARGS = {
    'clip': np_clip,
    'size': 256,
    'interpolation': 'bilinear'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 4, dtype family int — 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
tensor_normalize Ran
ran, honoured the contract we drafted
bruceyo/V-PETL/datasets/kinetics.py
pointer only (licence: MIT); the stored copy get_code serves, from kangsankim07/videoicl, is pointer only (licence: NONE) · get_code("14b3b13b2e7ee844")
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 kangsankim07/videoicl, 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.

window_partition Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bruceyo/V-PETL/models/video_swin_transformer_patt.py
pointer only (licence: MIT); the stored copy get_code serves, from shiqiyu/opengait, is pointer only (licence: NONE) · get_code("c33b3de22ecbd051")
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 shiqiyu/opengait, 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.

window_reverse Ran
ran; our contract draft was wrong, not the code
bruceyo/V-PETL/models/video_swin_transformer_patt.py
pointer only (licence: MIT); the stored copy get_code serves, from shiqiyu/opengait, is pointer only (licence: NONE) · get_code("ce51a79a418f8860")
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 shiqiyu/opengait, 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.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 12 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 0 the same month, 11 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
create_logger 35 other papers
9 dated earlier, 25 later, 1 same year, order not determinable
2104.13840 (2021-04)
Twins: Revisiting the Design of Spatial Attention in Vision Transformers
2106.03650 (2021-06)
Shuffle Transformer: Rethinking Spatial Shuffle for Vision Transformer
2106.03746 (2021-06)
Efficient Training of Visual Transformers with Small Datasets
and 32 more not named here
window_reverse 16 other papers
5 dated earlier, 11 later · 5 only as an assembly brick
2003.12039 (2020-03)
RAFT: Recurrent All-Pairs Field Transforms for Optical Flow
2106.13230 (2021-06)
Video Swin Transformer
2111.14821 (2021-11)
End-to-End Referring Video Object Segmentation with Multimodal Transformers
and 13 more not named here
build_transform 15 other papers
2 dated earlier, 13 later
2205.03892 (2022-05)
ConvMAE: Masked Convolution Meets Masked Autoencoders
2206.07699 (2022-06)
Write and Paint: Generative Vision-Language Models are Unified Modal Learners
2212.05677 (2022-12)
Masked autoencoders are effective solution to transformer data-hungry
and 12 more not named here
tensor_normalize 15 other papers
15 later
2211.09120 (2022-11)
AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning with Masked Autoencoders
2307.02227 (2023-07)
MAE-DFER: Efficient Masked Autoencoder for Self-supervised Dynamic Facial Expression Recognition
2308.04549 (2023-08)
Prune Spatio-temporal Tokens by Semantic-aware Temporal Accumulation
and 12 more not named here
window_partition 15 other papers
6 dated earlier, 9 later · 2 only as an assembly brick
2106.13230 (2021-06)
Video Swin Transformer
2111.13300 (2021-11)
A Robust Volumetric Transformer for Accurate 3D Tumor Segmentation
2111.14821 (2021-11)
End-to-End Referring Video Object Segmentation with Multimodal Transformers
and 12 more not named here
build_dataset 14 other papers
1 dated earlier, 13 later
2205.03892 (2022-05)
ConvMAE: Masked Convolution Meets Masked Autoencoders
aaai_25300 (2023 (year only))
Exploring Stochastic Autoregressive Image Modeling for Visual Representation
2303.07598 (2023-03)
AdPE: Adversarial Positional Embeddings for Pretraining Vision Transformers via MAE+
and 11 more not named here
crop_clip 13 other papers
2 dated earlier, 11 later
2112.02815 (2021-12)
Make It Move: Controllable Image-to-Video Generation with Text Descriptions
2209.10340 (2022-09)
FNeVR: Neural Volume Rendering for Face Animation
2211.09120 (2022-11)
AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning with Masked Autoencoders
and 10 more not named here
get_resize_sizes 10 other papers
10 later
2211.09120 (2022-11)
AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning with Masked Autoencoders
2307.02227 (2023-07)
MAE-DFER: Efficient Masked Autoencoder for Self-supervised Dynamic Facial Expression Recognition
2308.04549 (2023-08)
Prune Spatio-temporal Tokens by Semantic-aware Temporal Accumulation
and 7 more not named here
resize_clip 9 other papers
9 later
2211.09120 (2022-11)
AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning with Masked Autoencoders
2307.02227 (2023-07)
MAE-DFER: Efficient Masked Autoencoder for Self-supervised Dynamic Facial Expression Recognition
2308.04549 (2023-08)
Prune Spatio-temporal Tokens by Semantic-aware Temporal Accumulation
and 6 more not named here
crop_boxes 4 other papers
2 dated earlier, 1 later, 1 same year, order not determinable
Shi_Multi-Modal_Multi-Action_Video_Recognition_ICCV_2021_paper (2021 (year only))
Multi-Modal Multi-Action Video Recognition
136640364 (2022 (year only))
Continual 3D Convolutional Neural Networks for Real-time Processing of Videos
2206.13559 (2022-06)
ST-Adapter: Parameter-Efficient Image-to-Video Transfer Learning
and 1 more not named here
random_short_side_scale_jitter 3 other papers
1 dated earlier, 1 later, 1 same year, order not determinable
Shi_Multi-Modal_Multi-Action_Video_Recognition_ICCV_2021_paper (2021 (year only))
Multi-Modal Multi-Action Video Recognition
136640364 (2022 (year only))
Continual 3D Convolutional Neural Networks for Real-time Processing of Videos
2410.00263 (2024-10)
Procedure-Aware Surgical Video-language Pretraining with Hierarchical Knowledge Augmentation
random_crop 2 other papers
1 dated earlier, 1 same year, order not determinable
Shi_Multi-Modal_Multi-Action_Video_Recognition_ICCV_2021_paper (2021 (year only))
Multi-Modal Multi-Action Video Recognition
136640364 (2022 (year only))
Continual 3D Convolutional Neural Networks for Real-time Processing of Videos

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

Parameter efficient transfer learning (PETL) aims at making good use of the representation knowledge in the pre-trained large models by fine-tuning a small number of parameters. Recently, taking inspiration from the natural language processing (NLP) domain, popular PETL techniques such as prompt-tuning and Adapter have also been successfully applied to the vision domain. However, prefix-tuning remains under-explored for vision tasks. In this work, we intend to adapt large vision models (LVMs) to downstream tasks with a good parameter-accuracy trade-off. Towards this goal, we propose a framework with a unified view of PETL called visual-PETL (V-PETL) to investigate the effects of different PETL techniques, data scales of downstream domains, positions of trainable parameters, and other aspects affecting the trade-off. Specifically, we analyze the positional importance of trainable parameters and differences between NLP and vision tasks in terms of data structures and pre-training mechanisms while implementing various PETL techniques, especially for the under-explored prefix-tuning technique. Based on a comprehensive understanding of the differences between NLP and vision data, we propose a new variation of the prefix-tuning module called parallel attention (PATT) for vision downstream tasks. An extensive empirical analysis on vision tasks via different frozen LVMs has been carried and the findings show that the proposed PATT can effectively contribute to other PETL techniques. An effective scheme Swin-BAPAT derived from the proposed V-PETL framework achieves significantly better performance than the state-of-the-art AdaptFormer-Swin with slightly more parameters and outperforms full-tuning with far fewer parameters. Code and data are available at: https://github.com/bruceyo/V-PETL.

For agents

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

get_harvested_code_for_paper("2210.00788")
get_code_for_paper("2210.00788")
have("2210.00788")

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

curl https://syntology.ai/api/ran/2210.00788.json

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

[![Syntology run record](https://syntology.ai/api/ran/2210.00788.svg)](https://syntology.ai/paper/2210.00788)

Connect an agent — have() is free.