Despite achieving remarkable performance on various vision-language tasks, Transformer-based pretrained vision-language models (VLMs) still suffer from efficiency issues arising from long inputs and numerous parameters, limiting their real-world applications. However, the huge computation is redundant for most samples and the degree of redundancy and the respective components vary significantly depending on tasks and input instances. In this work, we propose an adaptive acceleration method SmartTrim for VLMs, which adjusts the inference overhead based on the complexity of instances. Specifically, SmartTrim incorporates lightweight trimming modules into the backbone to perform task-specific pruning on redundant inputs and parameters, without the need for additional pre-training or data augmentation. Since visual and textual representations complement each other in VLMs, we propose to leverage cross-modal interaction information to provide more critical semantic guidance for identifying redundant parts. Meanwhile, we introduce a self-distillation strategy that encourages the trimmed model to be consistent with the full-capacity model, which yields further performance gains. Experimental results demonstrate that SmartTrim significantly reduces the computation overhead (2-3 times) of various VLMs with comparable performance (only a 1-2% degradation) on various vision-language tasks. Compared to previous acceleration methods, SmartTrim attains a better efficiency-performance trade-off, demonstrating great potential for application in resource-constrained scenarios.
相關內容
Adapting pretrained language models to novel domains, such as clinical applications, traditionally involves retraining their entire set of parameters. However, this approach is increasingly proven to be impractical owing to the substantial computational requirements associated with training such large language models. To address this issue, Parameter-Efficient Fine-Tuning (PEFT) techniques offer a viable solution by selectively fine-tuning a small subset of additional parameters, significantly reducing the computational requirements for domain adaptation. In this study, we propose Clinical LLaMA-LoRA, a PEFT adapter layer built upon the open-sourced LLaMA model. Clinical LLaMA-LoRA is trained using clinical notes obtained from the MIMIC-IV database, thereby creating a specialised adapter designed for the clinical domain. Additionally, we propose a two-step PEFT framework which fuses Clinical LLaMA-LoRA with Downstream LLaMA-LoRA, another PEFT adapter specialised for downstream tasks. We evaluate this framework on multiple clinical outcome prediction datasets, comparing it to clinically trained language models. Our proposed framework achieves a state-of-the-art AUROC score averaged across all clinical downstream tasks. We observe substantial improvements of 6-9% AUROC score in the large-scale multilabel classification tasks, such as diagnoses and procedures classification.
Large-scale pre-trained models have been remarkably successful in resolving downstream tasks. Nonetheless, deploying these models on low-capability devices still requires an effective approach, such as model pruning. However, pruning the model from scratch can pose a practical challenge given the limited resources of each downstream task or device. To tackle this issue, we present a scalable one-shot pruning method that leverages pruned knowledge of similar tasks to extract a sub-network from the pre-trained model for a new task. Specifically, we create a score mask using the pruned models of similar tasks to identify task-specific filters/nodes in the pre-trained model for the new task. Based on this mask, we conduct a single round of pruning to extract a suitably-sized sub-network that can quickly adapt to the new task with only a few training iterations. Our experimental analysis demonstrates the effectiveness of the proposed method on the convolutional neural networks (CNNs) and vision transformers (ViT) with various datasets. The proposed method consistently outperforms popular pruning baseline methods in terms of accuracy and efficiency when dealing with diverse downstream tasks with different memory constraints.
Leveraging the low-power, event-driven computation and the inherent temporal dynamics, spiking neural networks (SNNs) are potentially ideal solutions for processing dynamic and asynchronous signals from event-based sensors. However, due to the challenges in training and the restrictions in architectural design, there are limited examples of competitive SNNs in the realm of event-based dense prediction when compared to artificial neural networks (ANNs). In this paper, we present an efficient spiking encoder-decoder network designed for large-scale event-based semantic segmentation tasks. This is achieved by optimizing the encoder using a hierarchical search method. To enhance learning from dynamic event streams, we harness the inherent adaptive threshold of spiking neurons to modulate network activation. Moreover, we introduce a dual-path Spiking Spatially-Adaptive Modulation (SSAM) block, specifically designed to enhance the representation of sparse events, thereby considerably improving network performance. Our proposed network achieves a 72.57% mean intersection over union (MIoU) on the DDD17 dataset and a 57.22% MIoU on the recently introduced, larger DSEC-Semantic dataset. This performance surpasses the current state-of-the-art ANNs by 4%, whilst consuming significantly less computational resources. To the best of our knowledge, this is the first study demonstrating SNNs outperforming ANNs in demanding event-based semantic segmentation tasks, thereby establishing the vast potential of SNNs in the field of event-based vision. Our source code will be made publicly accessible.
Current large language models can perform reasonably well on complex tasks that require step-by-step reasoning with few-shot learning. Are these models applying reasoning skills they have learnt during pre-training and reason outside of their training context, or are they simply memorizing their training corpus at finer granularity and have learnt to better understand their context? To tease apart these possibilities, we introduce ALERT, a benchmark and suite of analyses for assessing language models' reasoning ability comparing pre-trained and finetuned models on complex tasks that require reasoning skills to solve. ALERT provides a test bed to asses any language model on fine-grained reasoning skills, which spans over 20 datasets and covers 10 different reasoning skills. We leverage ALERT to further investigate the role of finetuning. With extensive empirical analysis we find that language models learn more reasoning skills such as textual entailment, abductive reasoning, and analogical reasoning during finetuning stage compared to pretraining state. We also find that when language models are finetuned they tend to overfit to the prompt template, which hurts the robustness of models causing generalization problems.
With the emergence of large pre-trained vison-language model like CLIP, transferable representations can be adapted to a wide range of downstream tasks via prompt tuning. Prompt tuning tries to probe the beneficial information for downstream tasks from the general knowledge stored in the pre-trained model. A recently proposed method named Context Optimization (CoOp) introduces a set of learnable vectors as text prompt from the language side. However, tuning the text prompt alone can only adjust the synthesized "classifier", while the computed visual features of the image encoder can not be affected , thus leading to sub-optimal solutions. In this paper, we propose a novel Dual-modality Prompt Tuning (DPT) paradigm through learning text and visual prompts simultaneously. To make the final image feature concentrate more on the target visual concept, a Class-Aware Visual Prompt Tuning (CAVPT) scheme is further proposed in our DPT, where the class-aware visual prompt is generated dynamically by performing the cross attention between text prompts features and image patch token embeddings to encode both the downstream task-related information and visual instance information. Extensive experimental results on 11 datasets demonstrate the effectiveness and generalization ability of the proposed method. Our code is available in //github.com/fanrena/DPT.
Vision language tasks, such as answering questions about or generating captions that describe an image, are difficult tasks for computers to perform. A relatively recent body of research has adapted the pretrained transformer architecture introduced in \citet{vaswani2017attention} to vision language modeling. Transformer models have greatly improved performance and versatility over previous vision language models. They do so by pretraining models on a large generic datasets and transferring their learning to new tasks with minor changes in architecture and parameter values. This type of transfer learning has become the standard modeling practice in both natural language processing and computer vision. Vision language transformers offer the promise of producing similar advancements in tasks which require both vision and language. In this paper, we provide a broad synthesis of the currently available research on vision language transformer models and offer some analysis of their strengths, limitations and some open questions that remain.
Large Language Models (LLMs) have significantly advanced natural language processing (NLP) with their impressive language understanding and generation capabilities. However, their performance may be suboptimal for long-tail or domain-specific tasks due to limited exposure to domain-specific knowledge and vocabulary. Additionally, the lack of transparency of most state-of-the-art (SOTA) LLMs, which can only be accessed via APIs, impedes further fine-tuning with custom data. Moreover, data privacy is a significant concern. To address these challenges, we propose the novel Parametric Knowledge Guiding (PKG) framework, which equips LLMs with a knowledge-guiding module to access relevant knowledge at runtime without altering the LLMs' parameters. Our PKG is based on open-source "white-box" small language models, allowing offline storage of any knowledge that LLMs require. We demonstrate that our PKG framework can enhance the performance of "black-box" LLMs on a range of long-tail and domain-specific downstream tasks requiring factual, tabular, medical, and multimodal knowledge.
With the rise of powerful pre-trained vision-language models like CLIP, it becomes essential to investigate ways to adapt these models to downstream datasets. A recently proposed method named Context Optimization (CoOp) introduces the concept of prompt learning -- a recent trend in NLP -- to the vision domain for adapting pre-trained vision-language models. Specifically, CoOp turns context words in a prompt into a set of learnable vectors and, with only a few labeled images for learning, can achieve huge improvements over intensively-tuned manual prompts. In our study we identify a critical problem of CoOp: the learned context is not generalizable to wider unseen classes within the same dataset, suggesting that CoOp overfits base classes observed during training. To address the problem, we propose Conditional Context Optimization (CoCoOp), which extends CoOp by further learning a lightweight neural network to generate for each image an input-conditional token (vector). Compared to CoOp's static prompts, our dynamic prompts adapt to each instance and are thus less sensitive to class shift. Extensive experiments show that CoCoOp generalizes much better than CoOp to unseen classes, even showing promising transferability beyond a single dataset; and yields stronger domain generalization performance as well. Code is available at //github.com/KaiyangZhou/CoOp.
Pre-trained Language Models (PLMs) have achieved great success in various Natural Language Processing (NLP) tasks under the pre-training and fine-tuning paradigm. With large quantities of parameters, PLMs are computation-intensive and resource-hungry. Hence, model pruning has been introduced to compress large-scale PLMs. However, most prior approaches only consider task-specific knowledge towards downstream tasks, but ignore the essential task-agnostic knowledge during pruning, which may cause catastrophic forgetting problem and lead to poor generalization ability. To maintain both task-agnostic and task-specific knowledge in our pruned model, we propose ContrAstive Pruning (CAP) under the paradigm of pre-training and fine-tuning. It is designed as a general framework, compatible with both structured and unstructured pruning. Unified in contrastive learning, CAP enables the pruned model to learn from the pre-trained model for task-agnostic knowledge, and fine-tuned model for task-specific knowledge. Besides, to better retain the performance of the pruned model, the snapshots (i.e., the intermediate models at each pruning iteration) also serve as effective supervisions for pruning. Our extensive experiments show that adopting CAP consistently yields significant improvements, especially in extremely high sparsity scenarios. With only 3% model parameters reserved (i.e., 97% sparsity), CAP successfully achieves 99.2% and 96.3% of the original BERT performance in QQP and MNLI tasks. In addition, our probing experiments demonstrate that the model pruned by CAP tends to achieve better generalization ability.
This paper shows that masked autoencoders (MAE) are scalable self-supervised learners for computer vision. Our MAE approach is simple: we mask random patches of the input image and reconstruct the missing pixels. It is based on two core designs. First, we develop an asymmetric encoder-decoder architecture, with an encoder that operates only on the visible subset of patches (without mask tokens), along with a lightweight decoder that reconstructs the original image from the latent representation and mask tokens. Second, we find that masking a high proportion of the input image, e.g., 75%, yields a nontrivial and meaningful self-supervisory task. Coupling these two designs enables us to train large models efficiently and effectively: we accelerate training (by 3x or more) and improve accuracy. Our scalable approach allows for learning high-capacity models that generalize well: e.g., a vanilla ViT-Huge model achieves the best accuracy (87.8%) among methods that use only ImageNet-1K data. Transfer performance in downstream tasks outperforms supervised pre-training and shows promising scaling behavior.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191