HyperTokens: Controlling Token Dynamics for Continual Video-Language Understanding

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Nguyen, Toan, Liu, Yang, De Melo, Celso, Salim, Flora D.
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914394938540032
author Nguyen, Toan
Liu, Yang
De Melo, Celso
Salim, Flora D.
author_facet Nguyen, Toan
Liu, Yang
De Melo, Celso
Salim, Flora D.
contents Continual VideoQA with multimodal LLMs is hindered by interference between tasks and the prohibitive cost of storing task-specific prompts. We introduce HyperTokens, a transformer-based token generator that produces fine-tuning tokens on demand, giving explicit control over prompt updates while keeping memory fixed. To suppress forgetting, we propose meta-inspired regularisers that look ahead to avoid task-specific sharp directions and anchor the evolving generator to prior tasks. We further connect our objective to sharpness-aware optimisation, providing insight into why it encourages flatter cross-task minima and improves retention. Beyond regularisation, HyperTokens exploits lightweight auxiliary multimodal supervision through shared generation weights; guided by a causal perspective, we design feasible objectives and surrogate mutual-information losses to regularise anti-causal cross-modal directions. Across two standard continual VideoQA benchmarks, HyperTokens achieves higher average accuracy with substantially lower forgetting. Finally, we introduce a challenging cross-modal ImageQA->VideoQA protocol and show that HyperTokens enables robust continual transfer in this setting.
format Preprint
id arxiv_https___arxiv_org_abs_2603_06662
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle HyperTokens: Controlling Token Dynamics for Continual Video-Language Understanding
Nguyen, Toan
Liu, Yang
De Melo, Celso
Salim, Flora D.
Computer Vision and Pattern Recognition
Machine Learning
Continual VideoQA with multimodal LLMs is hindered by interference between tasks and the prohibitive cost of storing task-specific prompts. We introduce HyperTokens, a transformer-based token generator that produces fine-tuning tokens on demand, giving explicit control over prompt updates while keeping memory fixed. To suppress forgetting, we propose meta-inspired regularisers that look ahead to avoid task-specific sharp directions and anchor the evolving generator to prior tasks. We further connect our objective to sharpness-aware optimisation, providing insight into why it encourages flatter cross-task minima and improves retention. Beyond regularisation, HyperTokens exploits lightweight auxiliary multimodal supervision through shared generation weights; guided by a causal perspective, we design feasible objectives and surrogate mutual-information losses to regularise anti-causal cross-modal directions. Across two standard continual VideoQA benchmarks, HyperTokens achieves higher average accuracy with substantially lower forgetting. Finally, we introduce a challenging cross-modal ImageQA->VideoQA protocol and show that HyperTokens enables robust continual transfer in this setting.
title HyperTokens: Controlling Token Dynamics for Continual Video-Language Understanding
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2603.06662