Black-Box Continual Learning for Vision-Language Models

• URL: http://arxiv.org/abs/2606.22999v1
• Date: Mon, 22 Jun 2026 08:14:55 GMT
• Title: Black-Box Continual Learning for Vision-Language Models
• Abstract summary: We introduce Black-CL, a more realistic benchmark for continuous learning.<n>Black-CL enforces three primary real-world challenges: weight and architecture inaccessibility, constrained computation, and task-agnostic inference.<n>Under this setting, we propose BETA, a simple yet effective baseline built on the key insight that solely optimizing textual prototypes can navigate the complexities of CL.
• Score: 36.35747221992653
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The rapid deployment of Vision-Language Models (VLMs) in dynamic environments necessitates the ability to learn continuously without forgetting. However, traditional continual learning (CL) settings often rely on white-box paradigms, which is increasingly invalidated by the shift toward cloud-hosted models. In this paper, we introduce Black-CL, a more realistic benchmark for VLMs that enforces three primary real-world challenges: weight and architecture inaccessibility, constrained computation, and task-agnostic inference. The learner can query only output embeddings or logits, with no gradient flow through or structural modification of the backbone. Current CL methodologies, which rely on backbone backpropagation or complex parameter expansion, are fundamentally incompatible with these constraints. Under this setting, we propose BETA, a simple yet effective baseline built on the key insight that solely optimizing textual prototypes can navigate the complexities of CL. BETA integrates three core components: Semantic Projection Accumulation (SPA) for incremental knowledge acquisition, Latent Distribution Replay (LDR) for anchoring the embedding space against catastrophic forgetting, and Test-Time Prototype Adaptation (TTPA) for dynamic, instance-aware boundary refinement. Extensive experiments across ten diverse datasets and various backbones demonstrate that BETA significantly outperforms existing black-box tuners. Remarkably, with only 0.05 M trainable parameters, a 180--3000$\times$ reduction compared to competitive methods, BETA achieves performance on par with or even exceeding white-box CL methods. We believe Black-CL and BETA provide a foundational framework for future advancements in continual learning and accelerates the transition of continual learning from academia to real-world systems.

Related papers

• AnE: Pushing the Reasoning Frontier of Multimodal LLMs via Anchor Evolution [61.593935260052795]
  Post-training viaSupervised Fine-Tuning (SFT) and Reinforcement Learning (RL) is crucial for enhancing reasoning in Multimodal Large Language Models (MLLMs)<n>Existing paradigms often reach a performance bottleneck due to the limitations of static data.<n>We propose Anchor Evolution (AnE), a new paradigm that integrates truth-anchored data curation and model evolution.
  arXiv  Detail & Related papers  (2026-05-25T08:26:34Z)
• HIFICL: High-Fidelity In-Context Learning for Multimodal Tasks [6.981025167004503]
  We introduce High-Fidelity In-Context Learning (HIFICL) to more faithfully model the ICL mechanism.<n>HIFICL consists of three key components: 1) a set of "virtual key-value pairs" to act as a learnable context, 2) a low-rank factorization for stable and regularized training, and 3) a simple end-to-end training objective.
  arXiv  Detail & Related papers  (2026-03-13T08:03:35Z)
• LLMBoost: Make Large Language Models Stronger with Boosting [45.63712975323876]
  We introduce LLMBoost, a novel ensemble fine-tuning framework for ensemble learning of LLMs.<n>Inspired by the boosting paradigm, LLMBoost incorporates three key innovations. First, a cross-model attention mechanism enables successor models to access and fuse hidden states from predecessors.<n>Second, a chain training paradigm progressively fine-tunes connected models with an error-suppression objective, ensuring that each model rectifies the mispredictions of its predecessor.<n>Third, a near-parallel inference paradigm design pipelines hidden states across models layer by layer, achieving inference efficiency approaching single-model decoding.
  arXiv  Detail & Related papers  (2025-12-26T07:16:41Z)
• BOFA: Bridge-Layer Orthogonal Low-Rank Fusion for CLIP-Based Class-Incremental Learning [84.56022893225422]
  Class-Incremental Learning (CIL) aims to continually learn new categories without forgetting previously acquired knowledge.<n>Applying vision-language models such as CLIP to CIL poses two major challenges: (1) adapting to downstream tasks often requires additional learnable modules, increasing model complexity and susceptibility to forgetting; and (2) while multi-modal representations offer complementary strengths, existing methods have yet to fully realize their potential in effectively integrating visual and textual modalities.
  arXiv  Detail & Related papers  (2025-11-14T15:51:40Z)
• EasySteer: A Unified Framework for High-Performance and Extensible LLM Steering [55.56674028743782]
  Large language model (LLM) steering has emerged as a promising paradigm for controlling model behavior at inference time.<n>We present EasySteer, a unified framework for high-performance, LLM steering built on vLLM.
  arXiv  Detail & Related papers  (2025-09-29T17:59:07Z)
• Scale, Don't Fine-tune: Guiding Multimodal LLMs for Efficient Visual Place Recognition at Test-Time [12.659582318581606]
  Current approaches, including Vision Foundation Models (VFMs) and Multimodal Large Language Models (MLLMs), enhance semantic understanding but suffer from high computational overhead and limited cross-domain transferability when fine-tuned.<n>We propose a novel framework employing Test-Time Scaling (TTS) that leverages vision-language alignment capabilities through Guidance-based methods for direct similarity scoring.<n>Our approach eliminates two-stage processing by employing structured prompts that generate length-controllable scoring outputs.
  arXiv  Detail & Related papers  (2025-09-02T09:25:13Z)
• Continual Learning for VLMs: A Survey and Taxonomy Beyond Forgetting [70.83781268763215]
  Vision-language models (VLMs) have achieved impressive performance across diverse multimodal tasks by leveraging large-scale pre-training.<n>VLMs face unique challenges such as cross-modal feature drift, parameter interference due to shared architectures, and zero-shot capability erosion.<n>This survey aims to serve as a comprehensive and diagnostic reference for researchers developing lifelong vision-language systems.
  arXiv  Detail & Related papers  (2025-08-06T09:03:10Z)
• Boosting All-in-One Image Restoration via Self-Improved Privilege Learning [72.35265021054471]
  Self-Improved Privilege Learning (SIPL) is a novel paradigm that overcomes limitations by extending the utility of privileged information (PI) beyond training into the inference stage.<n>Central to SIPL is Proxy Fusion, a lightweight module incorporating a learnable Privileged Dictionary.<n>Extensive experiments demonstrate that SIPL significantly advances the state-of-the-art on diverse all-in-one image restoration benchmarks.
  arXiv  Detail & Related papers  (2025-05-30T04:36:52Z)
• Improve Multi-Modal Embedding Learning via Explicit Hard Negative Gradient Amplifying [7.9925771591348065]
  Core contrastive learning paradigm remains largely unchanged from CLIP-style models to MLLMs.<n>In this work, we conduct a detailed analysis of the gradients of the info-NCE loss with respect to the query, positive, and negative samples.<n>We propose to explicitly amplify the gradients associated with hard negative samples, thereby encouraging the model to learn more discriminative embeddings.
  arXiv  Detail & Related papers  (2025-05-28T11:18:19Z)
• Boosting Continual Learning of Vision-Language Models via Mixture-of-Experts Adapters [65.15700861265432]
  We present a parameter-efficient continual learning framework to alleviate long-term forgetting in incremental learning with vision-language models. Our approach involves the dynamic expansion of a pre-trained CLIP model, through the integration of Mixture-of-Experts (MoE) adapters. To preserve the zero-shot recognition capability of vision-language models, we introduce a Distribution Discriminative Auto-Selector.
  arXiv  Detail & Related papers  (2024-03-18T08:00:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.