Papers with Code 아카이브

Safety Alignment

태스크의 하위 분류와 논문·코드·평가·기법·데이터셋을 탐색하세요.

전체 기법전체 데이터셋

하위 태스크

0

평가표

0

보존된 평가 기록

0

코드

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

https://github.com/AIM-Intelligence/SUDO

공식 코드

sudo rm -rf agentic_security

https://github.com/Isaac-theori/PCJailbreak

공식 코드

BiasJailbreak:Analyzing Ethical Biases and Jailbreak Vulnerabilities in Large Language Models

https://github.com/Madhur-1/SageLLMSafetyEval

공식 코드

$\texttt{SAGE}$: A Generic Framework for LLM Safety Evaluation

https://github.com/MurrayTom/SG-Bench

공식 코드

SG-Bench: Evaluating LLM Safety Generalization Across Diverse Tasks and Prompt Types

https://github.com/YangHao97/RedteamAudioLMMs

공식 코드

Audio Is the Achilles' Heel: Red Teaming Audio Large Multimodal Models

https://github.com/ZHZisZZ/emulated-disalignment

공식 코드

Emulated Disalignment: Safety Alignment for Large Language Models May Backfire!

https://github.com/ZHZisZZ/modpo

공식 코드

Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization

https://github.com/ZeroNLP/Eraser

공식 코드

Eraser: Jailbreaking Defense in Large Language Models via Unlearning Harmful Knowledge

https://github.com/adam-bjtu/system-2-alignment

공식 코드

Don't Command, Cultivate: An Exploratory Study of System-2 Alignment

https://github.com/aegis1863/xjailbreak

공식 코드

xJailbreak: Representation Space Guided Reinforcement Learning for Interpretable LLM Jailbreaking

https://github.com/ai45lab/x-boundary

공식 코드

X-Boundary: Establishing Exact Safety Boundary to Shield LLMs from Multi-Turn Jailbreaks without Compromising Usability

https://github.com/aladinD/SafeMERGE

공식 코드

SafeMERGE: Preserving Safety Alignment in Fine-Tuned Large Language Models via Selective Layer-Wise Model Merging

https://github.com/aoshuang92/splora

공식 코드

Safe Pruning LoRA: Robust Distance-Guided Pruning for Safety Alignment in Adaptation of LLMs

https://github.com/aryamanarora/bayesian-laws-icl

공식 코드

Bayesian scaling laws for in-context learning

https://github.com/averyma/pandas

공식 코드

PANDAS: Improving Many-shot Jailbreaking via Positive Affirmation, Negative Demonstration, and Adaptive Sampling

https://github.com/beijing-aisi/panda-guard

공식 코드

PandaGuard: Systematic Evaluation of LLM Safety against Jailbreaking Attacks

https://github.com/bmpixel/safety-residual-space

공식 코드

The Hidden Dimensions of LLM Alignment: A Multi-Dimensional Safety Analysis

https://github.com/carol-gutianle/latentsafety

공식 코드

Probing the Robustness of Large Language Models Safety to Latent Perturbations

https://github.com/cert-lab/safety-subspaces

공식 코드

Safety Subspaces are Not Distinct: A Fine-Tuning Case Study

https://github.com/chengcanwu/sap

공식 코드

Mitigating Fine-tuning Risks in LLMs via Safety-Aware Probing Optimization

https://github.com/circle-hit/mpo

공식 코드

MPO: Multilingual Safety Alignment via Reward Gap Optimization

https://github.com/clearloveclearlove/beat

공식 코드

Probe before You Talk: Towards Black-box Defense against Backdoor Unalignment for Large Language Models

https://github.com/csHuangfdu/SafeAligner

공식 코드

SafeAligner: Safety Alignment against Jailbreak Attacks via Response Disparity Guidance

https://github.com/cuhk-arise/psychobench

공식 코드

Who is ChatGPT? Benchmarking LLMs' Psychological Portrayal Using PsychoBench

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

공식 코드

Ferret: Faster and Effective Automated Red Teaming with Reward-Based Scoring Technique

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

공식 코드

Red-Teaming Large Language Models using Chain of Utterances for Safety-Alignment

https://github.com/declare-lab/safety-arithmetic

공식 코드

Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations

https://github.com/dsn-2024/dsn

공식 코드

Don't Say No: Jailbreaking LLM by Suppressing Refusal

https://github.com/echosechen/spa-vl-rlhf

공식 코드

SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model

https://github.com/ed-zh/parden

공식 코드

PARDEN, Can You Repeat That? Defending against Jailbreaks via Repetition

https://github.com/feiwang96/Data-Advisor

공식 코드

Data Advisor: Dynamic Data Curation for Safety Alignment of Large Language Models

https://github.com/fisht0ucher/trident

공식 코드

TRIDENT: Enhancing Large Language Model Safety with Tri-Dimensional Diversified Red-Teaming Data Synthesis

https://github.com/fyyfu/safetyalignnlp

공식 코드

Safety Alignment in NLP Tasks: Weakly Aligned Summarization as an In-Context Attack

https://github.com/gigabaozi/aed

공식 코드

Alignment-Enhanced Decoding:Defending via Token-Level Adaptive Refining of Probability Distributions

https://github.com/git-disl/awesome_llm-harmful-fine-tuning-papers

공식 코드

Harmful Fine-tuning Attacks and Defenses for Large Language Models: A Survey

https://github.com/git-disl/lisa

공식 코드

Lisa: Lazy Safety Alignment for Large Language Models against Harmful Fine-tuning Attack

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