Papers with Code 아카이브

Task Arithmetic

태스크의 하위 분류와 논문·코드·평가·기법·데이터셋을 탐색하세요.

전체 기법 전체 데이터셋

하위 태스크: 0
평가표: 0
보존된 평가 기록: 0

A task vector specifies a direction in the weight space of a pre-trained model, such that movement in that direction improves performance on the task. We build task vectors by subtracting the weights of a pre-trained model from the weights of the same model after fine-tuning on a task. We show that these task vectors can be modified and combined together through arithmetic operations such as negation and addition, and the behavior of the resulting model is steered accordingly.

코드

관련 논문과 평가 결과에 연결된 구현입니다. 출처에서 확인된 저자 구현에는 공식 코드 표시가 있습니다.

https://github.com/AntoAndGar/task_singular_vectors

공식 코드

Task Singular Vectors: Reducing Task Interference in Model Merging

https://github.com/EnnengYang/AdaMerging

공식 코드

AdaMerging: Adaptive Model Merging for Multi-Task Learning

https://github.com/declare-lab/resta

공식 코드

Language Models are Homer Simpson! Safety Re-Alignment of Fine-tuned Language Models through Task Arithmetic

https://github.com/deep-analysis-research/slta

공식 코드

Leveraging Submodule Linearity Enhances Task Arithmetic Performance in LLMs

https://github.com/detectivemb/task-arithmetic-for-zs-ir

공식 코드

Investigating Task Arithmetic for Zero-Shot Information Retrieval

https://github.com/fredzzhang/atlas

공식 코드

Knowledge Composition using Task Vectors with Learned Anisotropic Scaling

https://github.com/gortizji/tangent_task_arithmetic

공식 코드

Task Arithmetic in the Tangent Space: Improved Editing of Pre-Trained Models

https://github.com/hiyouga/llama-factory

공식 코드

Language Models are Homer Simpson! Safety Re-Alignment of Fine-tuned Language Models through Task Arithmetic

https://github.com/iurada/talos-task-arithmetic

공식 코드

Efficient Model Editing with Task-Localized Sparse Fine-tuning

https://github.com/kawakera-lab/crossmodeltransfer

공식 코드

Cross-Model Transfer of Task Vectors via Few-Shot Orthogonal Alignment

https://github.com/kyrie-23/linear_task_arithmetic

공식 코드

Fine-Tuning Attention Modules Only: Enhancing Weight Disentanglement in Task Arithmetic

https://github.com/kyrie-23/task_arithmetic_tangent

공식 코드

Fine-Tuning Attention Modules Only: Enhancing Weight Disentanglement in Task Arithmetic

https://github.com/mklimasz/language-arithmetic

공식 코드

No Train but Gain: Language Arithmetic for training-free Language Adapters enhancement

https://github.com/mlfoundations/task_vectors

공식 코드

Editing Models with Task Arithmetic

https://github.com/mmorafah/takfl

공식 코드

Towards Diverse Device Heterogeneous Federated Learning via Task Arithmetic Knowledge Integration

https://github.com/naver-ai/negmerge

공식 코드

NegMerge: Consensual Weight Negation for Strong Machine Unlearning

https://github.com/nik-dim/tall_masks

공식 코드

Localizing Task Information for Improved Model Merging and Compression

https://github.com/tanganke/peta

공식 코드

Parameter Efficient Multi-task Model Fusion with Partial Linearization

https://github.com/tanganke/subspace_fusion

공식 코드

Concrete Subspace Learning based Interference Elimination for Multi-task Model Fusion

https://github.com/tanganke/weight-ensembling_moe

공식 코드

Merging Multi-Task Models via Weight-Ensembling Mixture of Experts

https://github.com/thuccslab/mergeguard

공식 코드

Have You Merged My Model? On The Robustness of Large Language Model IP Protection Methods Against Model Merging

https://github.com/uiuctml/TaskVectorBasis

공식 코드

Efficient Model Editing with Task Vector Bases: A Theoretical Framework and Scalable Approach

https://github.com/uiuctml/localize-and-stitch

공식 코드

Localize-and-Stitch: Efficient Model Merging via Sparse Task Arithmetic

https://github.com/ukplab/iclr2024-model-merging

공식 코드

Model Merging by Uncertainty-Based Gradient Matching

https://github.com/yankd22/calm

공식 코드

CALM: Consensus-Aware Localized Merging for Multi-Task Learning

https://github.com/yifei-he/localize-and-stitch

공식 코드

Localize-and-Stitch: Efficient Model Merging via Sparse Task Arithmetic

https://github.com/ylsung/vl-merging

공식 코드

An Empirical Study of Multimodal Model Merging

https://github.com/declare-lab/red-instruct

Language Models are Homer Simpson! Safety Re-Alignment of Fine-tuned Language Models through Task Arithmetic

https://github.com/duguodong7/model-evolution

Localizing Task Information for Improved Model Merging and Compression

https://github.com/flowritecom/flow-merge

Editing Models with Task Arithmetic

https://github.com/idankinderman/fs_merge

Editing Models with Task Arithmetic

https://github.com/thennal10/fisher-nodes-merging

Editing Models with Task Arithmetic

https://github.com/yanyang19/ImPart

Editing Models with Task Arithmetic

https://github.com/zibojia/COCOCO

Editing Models with Task Arithmetic

출처: Papers with Code · CC BY-SA 4.0. 아카이브 자료와 새로 수집한 논문은 별도로 관리합니다.