Picture for Mario Fritz

Mario Fritz

FullCert: Deterministic End-to-End Certification for Training and Inference of Neural Networks

Jun 17, 2024
Viaarxiv icon

Dataset and Lessons Learned from the 2024 SaTML LLM Capture-the-Flag Competition

Add code
Jun 12, 2024
Viaarxiv icon

MultiMax: Sparse and Multi-Modal Attention Learning

Add code
Jun 04, 2024
Viaarxiv icon

Are you still on track!? Catching LLM Task Drift with Activations

Add code
Jun 02, 2024
Viaarxiv icon

Stealthy Imitation: Reward-guided Environment-free Policy Stealing

Add code
May 11, 2024
Viaarxiv icon

PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics

Add code
Apr 06, 2024
Figure 1 for PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics
Figure 2 for PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics
Figure 3 for PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics
Figure 4 for PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics
Viaarxiv icon

Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?

Add code
Mar 11, 2024
Figure 1 for Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?
Figure 2 for Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?
Figure 3 for Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?
Figure 4 for Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?
Viaarxiv icon

LLM Task Interference: An Initial Study on the Impact of Task-Switch in Conversational History

Add code
Feb 28, 2024
Figure 1 for LLM Task Interference: An Initial Study on the Impact of Task-Switch in Conversational History
Figure 2 for LLM Task Interference: An Initial Study on the Impact of Task-Switch in Conversational History
Figure 3 for LLM Task Interference: An Initial Study on the Impact of Task-Switch in Conversational History
Figure 4 for LLM Task Interference: An Initial Study on the Impact of Task-Switch in Conversational History
Viaarxiv icon

Exploring Value Biases: How LLMs Deviate Towards the Ideal

Feb 21, 2024
Figure 1 for Exploring Value Biases: How LLMs Deviate Towards the Ideal
Figure 2 for Exploring Value Biases: How LLMs Deviate Towards the Ideal
Figure 3 for Exploring Value Biases: How LLMs Deviate Towards the Ideal
Figure 4 for Exploring Value Biases: How LLMs Deviate Towards the Ideal
Viaarxiv icon

Adaptive Hierarchical Certification for Segmentation using Randomized Smoothing

Add code
Feb 13, 2024
Viaarxiv icon