paper-with-me

홈 › Papers

Partial Fine-Tuning: A Successor to Full Fine-Tuning for Vision Transformers

2023-12-25 · Peng Ye, Yongqi Huang, Chongjun Tu, Minglei Li, Tao Chen, Tong He, Wanli Ouyang

Fine-tuning pre-trained foundation models has gained significant popularity in various research fields. Existing methods for fine-tuning can be roughly divided into two categories, namely Parameter-Efficient Fine-Tuning and High-Performance Fine-Tuning. The former aims at improving efficiency, while the latter focuses on enhancing performance. Beyond these methods, we demonstrate that Partial Fine-Tuning can be an innovative and promising direction capable of concurrently enhancing both efficiency and accuracy. We first validate eight manually-defined partial fine-tuning strategies across kinds of datasets and vision transformer architectures, and find that some partial fine-tuning strategies (e.g., ffn only or attention only) can achieve better performance with fewer tuned parameters than full fine-tuning, and selecting appropriate layers is critical to partial fine-tuning. Thus, we propose a novel fine-tuned angle metric to guide the selection of appropriate layers for partial fine-tuning, making it flexible to be adapted to various scenarios for more practicable partial fine-tuning. Additionally, we show that partial fine-tuning can serve as a new dimension for Model Soups, improving both the model performance and generalization with fewer tuned parameters. Comprehensive experiments on a wide range of datasets and models validate the great potential of partial fine-tuning.

📄 PDF Abstract BibTeX arXiv:2312.15681

Code (0)

등록된 구현이 없습니다.

Tasks

parameter-efficient fine-tuning

Methods 이 논문이 사용한 방법론

Soups Compress an ensemble of models into a single one by averaging their weights (under certain pre-conditions).
Multi-Head Attention 설명 없음
Attention 설명 없음
Layer Normalization Unlike batch normalization, Layer Normalization directly estimates the normalization statistics from the summed inputs…
Residual Connection 설명 없음
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…

Similar Papers 제목 키워드 기반

Ensemble Successor Representations for Task Generalization in Offline-to-Online Reinforcement Learning

2024-05-12 · Changhong Wang, Xudong Yu, Chenjia Bai, Qiaosheng Zhang 외

In Reinforcement Learning (RL), training a policy from scratch with online experiences can be inefficient because of the difficulties in exploration. Recently, offline RL provides a promising solution by giving an initia…

Offline RLReinforcement Learning (RL)Representation Learning

APS: Active Pretraining with Successor Features

2021-08-31 · Hao liu, Pieter Abbeel

We introduce a new unsupervised pretraining objective for reinforcement learning. During the unsupervised reward-free pretraining phase, the agent maximizes mutual information between tasks and states induced by the poli…

Unsupervised Reinforcement Learning

A neurally plausible model learns successor representations in partially observable environments

2019-06-22 · NeurIPS 2019 12 · Eszter Vertes, Maneesh Sahani

Animals need to devise strategies to maximize returns while interacting with their environment based on incoming noisy sensory observations. Task-relevant states, such as the agent's location within an environment or the…

reinforcement-learningReinforcement LearningReinforcement Learning (RL)

Target Driven Visual Navigation with Hybrid Asynchronous Universal Successor Representations

2018-11-27 · Shamane Siriwardhana, Rivindu Weerasekera, Suranga Nanayakkara

Being able to navigate to a target with minimal supervision and prior knowledge is critical to creating human-like assistive agents. Prior work on map-based and map-less approaches have limited generalizability. In this …

NavigateVisual Navigation

Robust Few-Shot Vision-Language Model Adaptation

2025-06-05 · Hanxin Wang, Tian Liu, Shu Kong

Pretrained VLMs achieve strong performance on downstream tasks when adapted with just a few labeled examples. As the adapted models inevitably encounter out-of-distribution (OOD) test data that deviates from the in-distr…

Language ModelingLanguage ModellingmodelRetrieval