Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space Article Swipe

PDF

Mengqi Li , Changyao Tian , Renqiu Xia , Ning Liao , Weiwei Guo , Junchi Yan , Hongsheng Li , Jifeng Dai , Hao Li , Xue Yang ·

YOU? · · 2025 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2505.17011

We propose AdapTok, an adaptive temporal causal video tokenizer that can flexibly allocate tokens for different frames based on video content. AdapTok is equipped with a block-wise masking strategy that randomly drops tail tokens of each block during training, and a block causal scorer to predict the reconstruction quality of video frames using different numbers of tokens. During inference, an adaptive token allocation strategy based on integer linear programming is further proposed to adjust token usage given predicted scores. Such design allows for sample-wise, content-aware, and temporally dynamic token allocation under a controllable overall budget. Extensive experiments for video reconstruction and generation on UCF-101 and Kinetics-600 demonstrate the effectiveness of our approach. Without additional image data, AdapTok consistently improves reconstruction quality and generation performance under different token budgets, allowing for more scalable and token-efficient generative video modeling.

Related Topics

Learning Curve

Learning Theory (Education)

Practice (Learning Method)

Deep Learning

Learning Standards

Attention (Machine Learning)

Learning Disability

Decision Tree Learning

Adaptive Combat Rifle

Q-Learning

Higher Learning

Adaptive Immune System

Concepts

No concepts available.

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2505.17011
PDF: https://arxiv.org/pdf/2505.17011
OA Status: green
OpenAlex ID: https://openalex.org/W4415329721

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4415329721

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2505.17011

Digital Object Identifier
Title: Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2025

Year of publication
Publication date: 2025-05-22

Full publication date if available
Authors: Mengqi Li, Changyao Tian, Renqiu Xia, Ning Liao, Weiwei Guo, Junchi Yan, Hongsheng Li, Jifeng Dai, Hao Li, Xue Yang

List of authors in order
Landing page: https://arxiv.org/abs/2505.17011

Publisher landing page
PDF URL: https://arxiv.org/pdf/2505.17011

Direct link to full text PDF
Open access: Yes

Whether a free full text is available
OA status: green

Open access status per OpenAlex
OA URL: https://arxiv.org/pdf/2505.17011

Direct OA link when available
Cited by: 0

Total citation count in OpenAlex

Full payload

id	https://openalex.org/W4415329721
doi	https://doi.org/10.48550/arxiv.2505.17011
ids.doi	https://doi.org/10.48550/arxiv.2505.17011
ids.openalex	https://openalex.org/W4415329721
fwci	0.0
type	preprint
title	Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T10775
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.9984999895095825
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/1707
topics[0].subfield.display_name	Computer Vision and Pattern Recognition
topics[0].display_name	Generative Adversarial Networks and Image Synthesis
topics[1].id	https://openalex.org/T11439
topics[1].field.id	https://openalex.org/fields/17
topics[1].field.display_name	Computer Science
topics[1].score	0.9779999852180481
topics[1].domain.id	https://openalex.org/domains/3
topics[1].domain.display_name	Physical Sciences
topics[1].subfield.id	https://openalex.org/subfields/1707
topics[1].subfield.display_name	Computer Vision and Pattern Recognition
topics[1].display_name	Video Analysis and Summarization
topics[2].id	https://openalex.org/T11105
topics[2].field.id	https://openalex.org/fields/17
topics[2].field.display_name	Computer Science
topics[2].score	0.9731000065803528
topics[2].domain.id	https://openalex.org/domains/3
topics[2].domain.display_name	Physical Sciences
topics[2].subfield.id	https://openalex.org/subfields/1707
topics[2].subfield.display_name	Computer Vision and Pattern Recognition
topics[2].display_name	Advanced Image Processing Techniques
is_xpac	False
apc_list
apc_paid
language	en
locations[0].id	pmh:oai:arXiv.org:2505.17011
locations[0].is_oa	True
locations[0].source.id	https://openalex.org/S4306400194
locations[0].source.issn
locations[0].source.type	repository
locations[0].source.is_oa	True
locations[0].source.issn_l
locations[0].source.is_core	False
locations[0].source.is_in_doaj	False
locations[0].source.display_name	arXiv (Cornell University)
locations[0].source.host_organization	https://openalex.org/I205783295
locations[0].source.host_organization_name	Cornell University
locations[0].source.host_organization_lineage	https://openalex.org/I205783295
locations[0].license
locations[0].pdf_url	https://arxiv.org/pdf/2505.17011
locations[0].version	submittedVersion
locations[0].raw_type	text
locations[0].license_id
locations[0].is_accepted	False
locations[0].is_published	False
locations[0].raw_source_name
locations[0].landing_page_url	http://arxiv.org/abs/2505.17011
locations[1].id	doi:10.48550/arxiv.2505.17011
locations[1].is_oa	True
locations[1].source.id	https://openalex.org/S4306400194
locations[1].source.issn
locations[1].source.type	repository
locations[1].source.is_oa	True
locations[1].source.issn_l
locations[1].source.is_core	False
locations[1].source.is_in_doaj	False
locations[1].source.display_name	arXiv (Cornell University)
locations[1].source.host_organization	https://openalex.org/I205783295
locations[1].source.host_organization_name	Cornell University
locations[1].source.host_organization_lineage	https://openalex.org/I205783295
locations[1].license
locations[1].pdf_url
locations[1].version
locations[1].raw_type	article
locations[1].license_id
locations[1].is_accepted	False
locations[1].is_published
locations[1].raw_source_name
locations[1].landing_page_url	https://doi.org/10.48550/arxiv.2505.17011
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5100662610
authorships[0].author.orcid	https://orcid.org/0000-0002-9952-3690
authorships[0].author.display_name	Mengqi Li
authorships[0].author_position	first
authorships[0].raw_author_name	Li, Yan
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5087792832
authorships[1].author.orcid	https://orcid.org/0000-0002-3285-4671
authorships[1].author.display_name	Changyao Tian
authorships[1].author_position	middle
authorships[1].raw_author_name	Tian, Changyao
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5113021225
authorships[2].author.orcid
authorships[2].author.display_name	Renqiu Xia
authorships[2].author_position	middle
authorships[2].raw_author_name	Xia, Renqiu
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5111696111
authorships[3].author.orcid	https://orcid.org/0000-0002-3764-2555
authorships[3].author.display_name	Ning Liao
authorships[3].author_position	middle
authorships[3].raw_author_name	Liao, Ning
authorships[3].is_corresponding	False
authorships[4].author.id	https://openalex.org/A5027639981
authorships[4].author.orcid	https://orcid.org/0000-0001-5037-0972
authorships[4].author.display_name	Weiwei Guo
authorships[4].author_position	middle
authorships[4].raw_author_name	Guo, Weiwei
authorships[4].is_corresponding	False
authorships[5].author.id	https://openalex.org/A5087158377
authorships[5].author.orcid	https://orcid.org/0000-0001-9639-7679
authorships[5].author.display_name	Junchi Yan
authorships[5].author_position	middle
authorships[5].raw_author_name	Yan, Junchi
authorships[5].is_corresponding	False
authorships[6].author.id	https://openalex.org/A5100732450
authorships[6].author.orcid	https://orcid.org/0000-0002-2664-7975
authorships[6].author.display_name	Hongsheng Li
authorships[6].author_position	middle
authorships[6].raw_author_name	Li, Hongsheng
authorships[6].is_corresponding	False
authorships[7].author.id	https://openalex.org/A5026944066
authorships[7].author.orcid	https://orcid.org/0000-0002-6785-0785
authorships[7].author.display_name	Jifeng Dai
authorships[7].author_position	middle
authorships[7].raw_author_name	Dai, Jifeng
authorships[7].is_corresponding	False
authorships[8].author.id	https://openalex.org/A5100348486
authorships[8].author.orcid	https://orcid.org/0000-0001-6861-9430
authorships[8].author.display_name	Hao Li
authorships[8].author_position	middle
authorships[8].raw_author_name	Li, Hao
authorships[8].is_corresponding	False
authorships[9].author.id	https://openalex.org/A5101548928
authorships[9].author.orcid	https://orcid.org/0000-0001-6053-8855
authorships[9].author.display_name	Xue Yang
authorships[9].author_position	last
authorships[9].raw_author_name	Yang, Xue
authorships[9].is_corresponding	False
has_content.pdf	True
has_content.grobid_xml	True
is_paratext	False
open_access.is_oa	True
open_access.oa_url	https://arxiv.org/pdf/2505.17011
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-19T00:00:00
display_name	Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T10775
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.9984999895095825
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1707
primary_topic.subfield.display_name	Computer Vision and Pattern Recognition
primary_topic.display_name	Generative Adversarial Networks and Image Synthesis
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2505.17011
best_oa_location.is_oa	True
best_oa_location.source.id	https://openalex.org/S4306400194
best_oa_location.source.issn
best_oa_location.source.type	repository
best_oa_location.source.is_oa	True
best_oa_location.source.issn_l
best_oa_location.source.is_core	False
best_oa_location.source.is_in_doaj	False
best_oa_location.source.display_name	arXiv (Cornell University)
best_oa_location.source.host_organization	https://openalex.org/I205783295
best_oa_location.source.host_organization_name	Cornell University
best_oa_location.source.host_organization_lineage	https://openalex.org/I205783295
best_oa_location.license
best_oa_location.pdf_url	https://arxiv.org/pdf/2505.17011
best_oa_location.version	submittedVersion
best_oa_location.raw_type	text
best_oa_location.license_id
best_oa_location.is_accepted	False
best_oa_location.is_published	False
best_oa_location.raw_source_name
best_oa_location.landing_page_url	http://arxiv.org/abs/2505.17011
primary_location.id	pmh:oai:arXiv.org:2505.17011
primary_location.is_oa	True
primary_location.source.id	https://openalex.org/S4306400194
primary_location.source.issn
primary_location.source.type	repository
primary_location.source.is_oa	True
primary_location.source.issn_l
primary_location.source.is_core	False
primary_location.source.is_in_doaj	False
primary_location.source.display_name	arXiv (Cornell University)
primary_location.source.host_organization	https://openalex.org/I205783295
primary_location.source.host_organization_name	Cornell University
primary_location.source.host_organization_lineage	https://openalex.org/I205783295
primary_location.license
primary_location.pdf_url	https://arxiv.org/pdf/2505.17011
primary_location.version	submittedVersion
primary_location.raw_type	text
primary_location.license_id
primary_location.is_accepted	False
primary_location.is_published	False
primary_location.raw_source_name
primary_location.landing_page_url	http://arxiv.org/abs/2505.17011
publication_date	2025-05-22
publication_year	2025
referenced_works_count	0
abstract_inverted_index.a	25, 40, 91
abstract_inverted_index.We	0
abstract_inverted_index.an	3, 59
abstract_inverted_index.is	22, 69
abstract_inverted_index.of	34, 49, 55, 109
abstract_inverted_index.on	18, 65, 102
abstract_inverted_index.to	44, 72
abstract_inverted_index.and	39, 85, 100, 104, 121, 132
abstract_inverted_index.can	10
abstract_inverted_index.for	14, 82, 97, 129
abstract_inverted_index.our	110
abstract_inverted_index.the	46, 107
abstract_inverted_index.Such	79
abstract_inverted_index.each	35
abstract_inverted_index.more	130
abstract_inverted_index.tail	32
abstract_inverted_index.that	9, 29
abstract_inverted_index.with	24
abstract_inverted_index.based	17, 64
abstract_inverted_index.block	36, 41
abstract_inverted_index.data,	115
abstract_inverted_index.drops	31
abstract_inverted_index.given	76
abstract_inverted_index.image	114
abstract_inverted_index.token	61, 74, 88, 126
abstract_inverted_index.under	90, 124
abstract_inverted_index.usage	75
abstract_inverted_index.using	52
abstract_inverted_index.video	7, 19, 50, 98, 135
abstract_inverted_index.During	57
abstract_inverted_index.adjust	73
abstract_inverted_index.allows	81
abstract_inverted_index.causal	6, 42
abstract_inverted_index.design	80
abstract_inverted_index.during	37
abstract_inverted_index.frames	16, 51
abstract_inverted_index.linear	67
abstract_inverted_index.scorer	43
abstract_inverted_index.tokens	13, 33
abstract_inverted_index.AdapTok	21, 116
abstract_inverted_index.UCF-101	103
abstract_inverted_index.Without	112
abstract_inverted_index.budget.	94
abstract_inverted_index.dynamic	87
abstract_inverted_index.further	70
abstract_inverted_index.integer	66
abstract_inverted_index.masking	27
abstract_inverted_index.numbers	54
abstract_inverted_index.overall	93
abstract_inverted_index.predict	45
abstract_inverted_index.propose	1
abstract_inverted_index.quality	48, 120
abstract_inverted_index.scores.	78
abstract_inverted_index.tokens.	56
abstract_inverted_index.AdapTok,	2
abstract_inverted_index.adaptive	4, 60
abstract_inverted_index.allocate	12
abstract_inverted_index.allowing	128
abstract_inverted_index.budgets,	127
abstract_inverted_index.content.	20
abstract_inverted_index.equipped	23
abstract_inverted_index.flexibly	11
abstract_inverted_index.improves	118
abstract_inverted_index.proposed	71
abstract_inverted_index.randomly	30
abstract_inverted_index.scalable	131
abstract_inverted_index.strategy	28, 63
abstract_inverted_index.temporal	5
abstract_inverted_index.Extensive	95
abstract_inverted_index.approach.	111
abstract_inverted_index.different	15, 53, 125
abstract_inverted_index.modeling.	136
abstract_inverted_index.predicted	77
abstract_inverted_index.tokenizer	8
abstract_inverted_index.training,	38
abstract_inverted_index.additional	113
abstract_inverted_index.allocation	62, 89
abstract_inverted_index.block-wise	26
abstract_inverted_index.generation	101, 122
abstract_inverted_index.generative	134
abstract_inverted_index.inference,	58
abstract_inverted_index.temporally	86
abstract_inverted_index.demonstrate	106
abstract_inverted_index.experiments	96
abstract_inverted_index.performance	123
abstract_inverted_index.programming	68
abstract_inverted_index.Kinetics-600	105
abstract_inverted_index.consistently	117
abstract_inverted_index.controllable	92
abstract_inverted_index.sample-wise,	83
abstract_inverted_index.effectiveness	108
abstract_inverted_index.content-aware,	84
abstract_inverted_index.reconstruction	47, 99, 119
abstract_inverted_index.token-efficient	133
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	10
citation_normalized_percentile