PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference | Paper Citation Record

Observation 027b6d57-d2fe-414a-90a1-03ebb53350df · inbound

RED: Unleashing Token-Level Rewards from Holistic Feedback via Reward Redistribution cites this paper.

RED: Unleashing Token-Level Rewards from Holistic Feedback via Reward Redistribution PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-12T21:50:34.629915Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 542ffbf9-5bab-40b6-94f0-a5710bb8e41d · inbound

VLSBench: Unveiling Visual Leakage in Multimodal Safety cites this paper.

VLSBench: Unveiling Visual Leakage in Multimodal Safety PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-12T05:56:46.428134Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 59280202-38f0-494f-b3db-ed4b0f5e2bdb · inbound

Robust Multi-bit Text Watermark with LLM-based Paraphrasers cites this paper.

Robust Multi-bit Text Watermark with LLM-based Paraphrasers PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 2019

Resolution: unresolved; no resolver link, observed 2026-08-11T22:49:58.514761Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c07fd98b-a7fa-4a78-bc62-d6a2491ba213 · inbound

LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods cites this paper.

LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 100

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:08:36.843119Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 02230ff2-b5e2-494d-aec5-e82f8bc79f83 · inbound

Targeted Angular Reversal of Weights (TARS) for Knowledge Removal in Large Language Models cites this paper.

Targeted Angular Reversal of Weights (TARS) for Knowledge Removal in Large Language Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-11T16:14:25.014603Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b22ca459-5270-4b9e-9092-fe28562c7324 · inbound

Align Anything: Training All-Modality Models to Follow Instructions with Language Feedback cites this paper.

Align Anything: Training All-Modality Models to Follow Instructions with Language Feedback PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 48

Resolution: unresolved; no resolver link, observed 2026-08-11T11:09:13.150282Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1c2e0654-df77-4799-972b-b57bcfff8d00 · inbound

Multi-Objective Large Language Model Unlearning cites this paper.

Multi-Objective Large Language Model Unlearning PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-10T23:27:37.742460Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7447808b-12ec-4dbe-8aa2-aa4544fb5fcf · inbound

Stream Aligner: Efficient Sentence-Level Alignment via Distribution Induction cites this paper.

Stream Aligner: Efficient Sentence-Level Alignment via Distribution Induction PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-10T21:17:48.381225Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fa50e317-59f1-41fc-bd43-4e29d0173a98 · inbound

Gradient-Based Multi-Objective Deep Learning: Algorithms, Theories, Applications, and Beyond cites this paper.

Gradient-Based Multi-Objective Deep Learning: Algorithms, Theories, Applications, and Beyond PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 72

Resolution: unresolved; no resolver link, observed 2026-08-10T18:53:50.107993Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b12fa28a-d167-4fcf-aeb5-fa8ad8b9afc4 · inbound

Data-adaptive Safety Rules for Training Reward Models cites this paper.

Data-adaptive Safety Rules for Training Reward Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-10T14:25:13.146914Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e12b652f-af52-4be7-b387-036e5cfe24d3 · inbound

On Almost Surely Safe Alignment of Large Language Models at Inference-Time cites this paper.

On Almost Surely Safe Alignment of Large Language Models at Inference-Time PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 70

Resolution: unresolved; no resolver link, observed 2026-08-09T16:18:40.813673Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 44ab9950-d943-44e9-9e48-8df82a5721f4 · inbound

Vulnerability Mitigation for Safety-Aligned Language Models via Debiasing cites this paper.

Vulnerability Mitigation for Safety-Aligned Language Models via Debiasing PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-09T13:14:34.037586Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 53647d4d-ed38-470f-a8bb-4640aef25a66 · inbound

Safety Reasoning with Guidelines cites this paper.

Safety Reasoning with Guidelines PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-08T23:50:35.767760Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 238801f1-b654-41a6-a352-bebadbb6a97d · inbound

How Humans Help LLMs: Assessing and Incentivizing Human Preference Annotators cites this paper.

How Humans Help LLMs: Assessing and Incentivizing Human Preference Annotators PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 53

Resolution: verified exact; arxiv_id, observed 2026-05-23T03:57:29.606200Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation d882c0ce-848e-41dd-9599-3728f085f572 · inbound

AI Alignment at Your Discretion cites this paper.

AI Alignment at Your Discretion PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 54

Resolution: unresolved; no resolver link, observed 2026-08-08T16:14:57.329365Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 95db99d9-5da6-4d58-94bd-453783b2ad5a · inbound

Probing and Inducing Combinational Creativity in Vision-Language Models cites this paper.

Probing and Inducing Combinational Creativity in Vision-Language Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-16T12:18:52.479377Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d8c0ae58-70ef-46b8-a18d-15456ea83bfe · inbound

A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment cites this paper.

A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 265

Resolution: unresolved; no resolver link, observed 2026-08-16T11:24:12.809889Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2f93b59b-02a3-4d92-a9e1-683d7951790e · inbound

PARM: Multi-Objective Test-Time Alignment via Preference-Aware Autoregressive Reward Model cites this paper.

PARM: Multi-Objective Test-Time Alignment via Preference-Aware Autoregressive Reward Model PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 2023

Resolution: unresolved; no resolver link, observed 2026-08-15T23:52:01.523681Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 06262224-13c3-4f23-8a03-21b71a1463e5 · inbound

J1: Exploring Simple Test-Time Scaling for LLM-as-a-Judge cites this paper.

J1: Exploring Simple Test-Time Scaling for LLM-as-a-Judge PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-15T20:50:10.534659Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2de9dde4-b320-4095-864b-2187a14f885d · inbound

SLMEval: Entropy-Based Calibration for Human-Aligned Evaluation of Large Language Models cites this paper.

SLMEval: Entropy-Based Calibration for Human-Aligned Evaluation of Large Language Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-07T15:12:26.903646Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 795a9b5c-9ae4-401d-9fa3-edf03b080874 · inbound

MPO: Multilingual Safety Alignment via Reward Gap Optimization cites this paper.

MPO: Multilingual Safety Alignment via Reward Gap Optimization PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 34

Resolution: unresolved; no resolver link, observed 2026-08-07T14:57:27.885966Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e349f15f-85a8-4845-a1ec-db80e8b5f095 · inbound

MTSA: Multi-turn Safety Alignment for LLMs through Multi-round Red-teaming cites this paper.

MTSA: Multi-turn Safety Alignment for LLMs through Multi-round Red-teaming PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-07T15:09:58.475342Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fe891c49-9c68-4b27-a31d-636d3040cd64 · inbound

SweEval: Do LLMs Really Swear? A Safety Benchmark for Testing Limits for Enterprise Use cites this paper.

SweEval: Do LLMs Really Swear? A Safety Benchmark for Testing Limits for Enterprise Use PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-07T14:52:28.092591Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6390b21b-79ca-4333-85d3-e69482b152e3 · inbound

Incentivizing High-Quality Human Annotations with Golden Questions cites this paper.

Incentivizing High-Quality Human Annotations with Golden Questions PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 23

Resolution: verified exact; arxiv_id, observed 2026-05-19T13:42:19.389017Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 62fb4ef6-d5c8-4425-9bab-d36d4cfe9f14 · inbound

Beyond Safe Answers: A Benchmark for Evaluating True Risk Awareness in Large Reasoning Models cites this paper.

Beyond Safe Answers: A Benchmark for Evaluating True Risk Awareness in Large Reasoning Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 31

Resolution: unresolved; no resolver link, observed 2026-08-07T14:13:52.457692Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5f05f867-ff0e-4065-8f01-bf7651d187a3 · inbound

Token-level Accept or Reject: A Micro Alignment Approach for Large Language Models cites this paper.

Token-level Accept or Reject: A Micro Alignment Approach for Large Language Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-07T14:11:31.054317Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a158b6c6-fa85-4b43-9101-14ce2a31ae06 · inbound

Lifelong Safety Alignment for Language Models cites this paper.

Lifelong Safety Alignment for Language Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-07T14:00:05.888953Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 75620268-4462-4d24-b277-de0d52d9690e · inbound

Bounded Rationality for LLMs: Satisficing Alignment at Inference-Time cites this paper.

Bounded Rationality for LLMs: Satisficing Alignment at Inference-Time PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-07T12:44:20.174221Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation eefbd577-524a-4cc1-a8d0-3faa2c8147af · inbound

Jailbreak-R1: Exploring the Jailbreak Capabilities of LLMs via Reinforcement Learning cites this paper.

Jailbreak-R1: Exploring the Jailbreak Capabilities of LLMs via Reinforcement Learning PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-07T12:01:08.749164Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation eed00d0e-6f1d-4820-a85f-fc1adb3eada8 · inbound

RewardAnything: Generalizable Principle-Following Reward Models cites this paper.

RewardAnything: Generalizable Principle-Following Reward Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 32

Resolution: unresolved; no resolver link, observed 2026-08-07T11:04:06.823967Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation eda1e4c2-d2fb-449e-b7a7-bf446442c1ca · inbound

Relic: Enhancing Reward Model Generalization for Low-Resource Indic Languages with Few-Shot Examples cites this paper.

Relic: Enhancing Reward Model Generalization for Low-Resource Indic Languages with Few-Shot Examples PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-15T19:32:11.801146Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation dbe04a67-8c2d-4973-a02e-dd0938068495 · inbound

MetaCipher: A Time-Persistent and Universal Multi-Agent Framework for Cipher-Based Jailbreak Attacks for LLMs cites this paper.

MetaCipher: A Time-Persistent and Universal Multi-Agent Framework for Cipher-Based Jailbreak Attacks for LLMs PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-06T22:08:22.809009Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3f47598e-ec16-4990-aec8-02e509777422 · inbound

Gradient-Adaptive Policy Optimization: Towards Multi-Objective Alignment of Large Language Models cites this paper.

Gradient-Adaptive Policy Optimization: Towards Multi-Objective Alignment of Large Language Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-06T20:49:52.401761Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 42fe3188-5383-423d-b63f-299496c0b3a7 · inbound

Tiny Reward Models cites this paper.

Tiny Reward Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-06T17:48:07.137146Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a1f836f5-fb40-4bdb-9479-ea1315d2ea15 · inbound

HKGAI-V1: Towards Regional Sovereign Large Language Model for Hong Kong cites this paper.

HKGAI-V1: Towards Regional Sovereign Large Language Model for Hong Kong PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-06T17:38:16.221861Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bd353d43-b1bd-435a-ae9f-85c017eed29b · inbound

The Safety Gap Toolkit: Evaluating Hidden Dangers of Open-Source Models cites this paper.

The Safety Gap Toolkit: Evaluating Hidden Dangers of Open-Source Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-06T19:08:23.564199Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7cee2df6-4214-4911-bf5f-530698648edc · inbound

Frontier AI Risk Management Framework in Practice: A Risk Analysis Technical Report cites this paper.

Frontier AI Risk Management Framework in Practice: A Risk Analysis Technical Report PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-06T15:14:22.849329Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8a1e5778-804d-43e5-9f4f-b251132c5784 · inbound

Generative Model Unlearning: A Survey through Target Events, Unlearning Operators, and Evaluation Protocols cites this paper.

Generative Model Unlearning: A Survey through Target Events, Unlearning Operators, and Evaluation Protocols PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 97

Resolution: unresolved; no resolver link, observed 2026-08-06T13:54:39.777864Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation eff3e40d-afc2-4c4a-adf1-c989463ac2dc · inbound

Forewarned is Forearmed: Pre-Synthesizing Jailbreak-like Instructions to Enhance LLM Safety Guardrail to Potential Attacks cites this paper.

Forewarned is Forearmed: Pre-Synthesizing Jailbreak-like Instructions to Enhance LLM Safety Guardrail to Potential Attacks PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-05T15:19:28.700858Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c95821df-2fb0-4028-9f2c-4fb40de35681 · inbound

Beyond Prediction: Reinforcement Learning as the Defining Leap in Healthcare AI cites this paper.

Beyond Prediction: Reinforcement Learning as the Defining Leap in Healthcare AI PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 79

Resolution: unresolved; no resolver link, observed 2026-08-05T15:08:54.032086Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 223eb6c5-4b62-4bb9-b4aa-58793cd83de8 · inbound

The Realignment Problem: When Right becomes Wrong in LLMs cites this paper.

The Realignment Problem: When Right becomes Wrong in LLMs PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 10

Resolution: verified exact; arxiv_id, observed 2026-05-18T01:30:35.748934Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 25368b2e-73a0-462d-b0d2-a24ebb1d63cc · inbound

SCOUT: A Defense Against Data Poisoning Attacks in Fine-Tuned Language Models cites this paper.

SCOUT: A Defense Against Data Poisoning Attacks in Fine-Tuned Language Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 29

Resolution: verified exact; arxiv_id, observed 2026-05-16T23:28:40.673267Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation e537745e-1e07-4d3c-a064-b218c88a2be4 · inbound

Safety Alignment as Continual Learning: Mitigating the Alignment Tax via Orthogonal Gradient Projection cites this paper.

Safety Alignment as Continual Learning: Mitigating the Alignment Tax via Orthogonal Gradient Projection PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 14

Resolution: verified exact; arxiv_id, observed 2026-05-16T06:40:42.297455Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation be7a05a0-4339-4a04-b476-d341a4c317a7 · inbound

Hierarchical Semantic Correlation-Aware Masked Autoencoder for Unsupervised Audio-Visual Representation Learning cites this paper.

Hierarchical Semantic Correlation-Aware Masked Autoencoder for Unsupervised Audio-Visual Representation Learning PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 10

Resolution: unresolved; no resolver link, observed 2026-07-13T10:41:51.251736Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4e1ba725-ec50-426f-8a97-8b3346a4580b · inbound

FedDetox: Robust Federated SLM Alignment via On-Device Data Sanitization cites this paper.

FedDetox: Robust Federated SLM Alignment via On-Device Data Sanitization PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 25

Resolution: verified exact; arxiv_id, observed 2026-05-11T06:56:01.706661Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 36019f95-c876-420d-b347-bec1c4ad3951 · inbound

Principles Do Not Apply Themselves: A Hermeneutic Perspective on AI Alignment cites this paper.

Principles Do Not Apply Themselves: A Hermeneutic Perspective on AI Alignment PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 13

Resolution: verified exact; arxiv_id, observed 2026-05-11T10:41:02.849441Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation f16c24b6-74c0-48d8-8c32-627d49b8987d · inbound

Characterizing Model-Native Skills cites this paper.

Characterizing Model-Native Skills PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 99

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T05:56:11.648277Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 1116b0a4-0e5d-4143-9f40-f09da8eb3e57 · inbound

LLM Safety From Within: Detecting Harmful Content with Internal Representations cites this paper.

LLM Safety From Within: Detecting Harmful Content with Internal Representations PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 8

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T12:20:23.122827Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation a1430dbe-54c2-4812-8474-047c9a47b1cc · inbound

SafetyALFRED: Evaluating Safety-Conscious Planning of Multimodal Large Language Models cites this paper.

SafetyALFRED: Evaluating Safety-Conscious Planning of Multimodal Large Language Models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 22

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T02:22:20.648413Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 421aee01-67c4-4e2e-9c11-22bc5ba68250 · inbound

Disentangling Intent from Role: Adversarial Self-Play for Persona-Invariant Safety Alignment cites this paper.

Disentangling Intent from Role: Adversarial Self-Play for Persona-Invariant Safety Alignment PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 46

Resolution: verified exact; arxiv_id, observed 2026-05-11T16:21:07.450631Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 443f9f7a-6264-41a6-b089-a718eeda634d · inbound

Theoretical Limits of Language Model Alignment cites this paper.

Theoretical Limits of Language Model Alignment PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 28

Resolution: verified exact; arxiv_id, observed 2026-05-11T04:30:57.130357Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 92d2c7c5-7c5f-4519-a9b3-d91e03622f41 · inbound

GLiGuard: Schema-Conditioned Classification for LLM Safeguard cites this paper.

GLiGuard: Schema-Conditioned Classification for LLM Safeguard PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 5

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T03:20:55.235357Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation d4cf753a-c288-435a-a064-76e6dee5dbb9 · inbound

Structure from Strategic Interaction & Uncertainty: Risk Sensitive Games for Robust Preference Learning cites this paper.

Structure from Strategic Interaction & Uncertainty: Risk Sensitive Games for Robust Preference Learning PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 69

Resolution: metadata mismatch; arxiv_id, observed 2026-05-12T06:26:26.493049Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 55a935d3-6ded-46ec-9a05-70d7f56a358a · inbound

Structure from Strategic Interaction & Uncertainty: Risk Sensitive Games for Robust Preference Learning cites this paper.

Structure from Strategic Interaction & Uncertainty: Risk Sensitive Games for Robust Preference Learning PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 69

Resolution: metadata mismatch; arxiv_id, observed 2026-05-14T22:08:05.044304Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 14b214f3-f55b-43a4-8c5d-4fb4bae7ccc4 · inbound

Explaining and Breaking the Safety-Helpfulness Ceiling via Preference Dimensional Expansion cites this paper.

Explaining and Breaking the Safety-Helpfulness Ceiling via Preference Dimensional Expansion PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 22

Resolution: verified exact; arxiv_id, observed 2026-05-13T01:07:00.573759Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 79d9af60-23e5-4e2c-9a54-3f79909f8d3a · inbound

Explaining and Breaking the Safety-Helpfulness Ceiling via Preference Dimensional Expansion cites this paper.

Explaining and Breaking the Safety-Helpfulness Ceiling via Preference Dimensional Expansion PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 22

Resolution: verified exact; arxiv_id, observed 2026-05-14T21:12:58.932654Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 78a7f548-801c-4d12-93df-971461538fbb · inbound

Toward Stable Value Alignment: Introducing Independent Modules for Consistent Value Guidance cites this paper.

Toward Stable Value Alignment: Introducing Independent Modules for Consistent Value Guidance PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 81

Resolution: metadata mismatch; arxiv_id, observed 2026-05-13T07:07:27.645410Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation ee8bb0a1-922b-4ca0-b682-34ddbaad234b · inbound

Curriculum Learning for Safety Alignment cites this paper.

Curriculum Learning for Safety Alignment PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 10

Resolution: verified exact; arxiv_id, observed 2026-06-29T22:34:02.198090Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 8ef90606-1dae-45e7-b924-a55ef5ee26a2 · inbound

Opir: Efficient Multi-Task Safety Classification for Toxicity, Jailbreaks, Hate Speech, and Harmful Content cites this paper.

Opir: Efficient Multi-Task Safety Classification for Toxicity, Jailbreaks, Hate Speech, and Harmful Content PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 22

Resolution: verified exact; arxiv_id, observed 2026-06-29T09:13:15.988572Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 97042b30-7a37-4aa3-a47b-2d98b0242395 · inbound

Defending Jailbreak Attacks on Large Language Models via Manifold Trajectory Kinetics cites this paper.

Defending Jailbreak Attacks on Large Language Models via Manifold Trajectory Kinetics PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 90

Resolution: verified exact; arxiv_id, observed 2026-07-02T17:37:14.846238Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 4358c7bf-53a8-46c3-bbff-052a33ebadd0 · inbound

Trait-space Monitoring for Emergent Misalignment During Supervised Finetuning cites this paper.

Trait-space Monitoring for Emergent Misalignment During Supervised Finetuning PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 63

Resolution: verified exact; arxiv_id, observed 2026-07-01T20:56:13.867485Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 0a43df0a-b0bd-4264-972b-f9e371709a1e · inbound

When Behavioral Safety Evaluation Fails: A Representation-Level Perspective cites this paper.

When Behavioral Safety Evaluation Fails: A Representation-Level Perspective PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 17

Resolution: metadata mismatch; arxiv_id, observed 2026-07-02T20:57:23.032922Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation c595a4c3-30a6-4659-9233-e8b39098f680 · inbound

Multi-Objective Exploration and Preference Optimization via Mutual Information cites this paper.

Multi-Objective Exploration and Preference Optimization via Mutual Information PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 11

Resolution: unresolved; no resolver link, observed 2026-07-12T08:52:34.039089Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 601de794-7e48-4b62-9180-effb850bbf3b · inbound

OpenSafeIntent: Evaluating Intent-Calibrated Safe Completion Across Dual-Use Prompt Sets cites this paper.

OpenSafeIntent: Evaluating Intent-Calibrated Safe Completion Across Dual-Use Prompt Sets PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 23

Resolution: metadata mismatch; arxiv_id, observed 2026-07-03T14:48:32.496348Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation eb8b634a-2ac2-4c99-8aa1-543c4c1076f7 · inbound

Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarks cites this paper.

Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarks PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 276

Resolution: metadata mismatch; local_arxiv, observed 2026-07-10T15:47:23.388056Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation dd406a54-ffe1-4442-ad2e-102896795bda · inbound

Step-Level Preference Learning for Generative Agents in Social Simulations cites this paper.

Step-Level Preference Learning for Generative Agents in Social Simulations PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-02T02:00:25.508169Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9d5a050a-24bd-4eac-ac8d-78d296a66c50 · inbound

TRACE: Trajectory-Based Safety Patch Learning for LLM Post-Training Realignment cites this paper.

TRACE: Trajectory-Based Safety Patch Learning for LLM Post-Training Realignment PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-02T10:00:10.523132Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation aaeee69b-2017-43c0-a70b-38d6b7350112 · inbound

SCOPE and SCION: A Benchmark and an Auditable Reference Pipeline for Schema Induction and Fusion from Text cites this paper.

SCOPE and SCION: A Benchmark and an Auditable Reference Pipeline for Schema Induction and Fusion from Text PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-02T13:36:55.305100Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c5a4afa0-c396-49aa-b748-5bb4d4808ac4 · inbound

Visual Token Compression Enhances Robustness of MLLMs cites this paper.

Visual Token Compression Enhances Robustness of MLLMs PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 33

Resolution: unresolved; no resolver link, observed 2026-08-01T13:10:37.372979Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 354bbca7-6e30-44f8-8921-d118c8d1b3ca · inbound

When Do Task Vectors Interfere? Mapping the Validity Boundaries of Weight-Space Composition cites this paper.

When Do Task Vectors Interfere? Mapping the Validity Boundaries of Weight-Space Composition PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-11T16:35:24.771726Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 86642dfe-8e5c-417f-a708-d95cbcb5edf8 · inbound

Orientation, not magnitude: the causal structure of task-vector interference in merged language models cites this paper.

Orientation, not magnitude: the causal structure of task-vector interference in merged language models PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference

Reference 2023

Resolution: unresolved; no resolver link, observed 2026-08-16T00:35:14.975581Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Coverage vector

Reference resolution

Outbound references

Pith citing papers