Data of the "Leaky Thoughts: Large Reasoning Models Are Not Private Thinkers" paper
AI & ML interests
LLM, trustworthy AI, AI security, privacy, calibration, hallucination
Recent Activity
View all activity
Papers
Privacy Collapse: Benign Fine-Tuning Can Break Contextual Privacy in Language Models
Is Multilingual LLM Watermarking Truly Multilingual? A Simple Back-Translation Solution
NAACL 2025 Findings "Scaling Up Membership Inference: When and How Attacks Succeed on Large Language Models" https://arxiv.org/abs/2411.00154
List of research articles of Parameter Lab
-
Privacy Collapse: Benign Fine-Tuning Can Break Contextual Privacy in Language Models
Paper • 2601.15220 • Published • 8 -
Dr.LLM: Dynamic Layer Routing in LLMs
Paper • 2510.12773 • Published • 32 -
Is Multilingual LLM Watermarking Truly Multilingual? A Simple Back-Translation Solution
Paper • 2510.18019 • Published • 18 -
Leaky Thoughts: Large Reasoning Models Are Not Private Thinkers
Paper • 2506.15674 • Published • 2
Fine-tuned models for black-box LLM calibration, trained for "Apricot: Calibrating Large Language Models Using Their Generations Only" (ACL 2024)
-
parameterlab/apricot_binary_trivia_qa_deberta-v3-base_for_vicuna-7b-v1.5
Text Classification • 0.2B • Updated • 4 • 1 -
parameterlab/apricot_clustering_trivia_qa_deberta-v3-base_for_vicuna-7b-v1.5
Text Classification • 0.2B • Updated • 12 -
parameterlab/apricot_binary_coqa_deberta-v3-base_for_vicuna-7b-v1.5
Text Classification • 0.2B • Updated • 8 -
parameterlab/apricot_clustering_coqa_deberta-v3-base_for_vicuna-7b-v1.5
Text Classification • 0.2B • Updated • 10
Data of the "Leaky Thoughts: Large Reasoning Models Are Not Private Thinkers" paper
List of research articles of Parameter Lab
-
Privacy Collapse: Benign Fine-Tuning Can Break Contextual Privacy in Language Models
Paper • 2601.15220 • Published • 8 -
Dr.LLM: Dynamic Layer Routing in LLMs
Paper • 2510.12773 • Published • 32 -
Is Multilingual LLM Watermarking Truly Multilingual? A Simple Back-Translation Solution
Paper • 2510.18019 • Published • 18 -
Leaky Thoughts: Large Reasoning Models Are Not Private Thinkers
Paper • 2506.15674 • Published • 2
NAACL 2025 Findings "Scaling Up Membership Inference: When and How Attacks Succeed on Large Language Models" https://arxiv.org/abs/2411.00154
Fine-tuned models for black-box LLM calibration, trained for "Apricot: Calibrating Large Language Models Using Their Generations Only" (ACL 2024)
-
parameterlab/apricot_binary_trivia_qa_deberta-v3-base_for_vicuna-7b-v1.5
Text Classification • 0.2B • Updated • 4 • 1 -
parameterlab/apricot_clustering_trivia_qa_deberta-v3-base_for_vicuna-7b-v1.5
Text Classification • 0.2B • Updated • 12 -
parameterlab/apricot_binary_coqa_deberta-v3-base_for_vicuna-7b-v1.5
Text Classification • 0.2B • Updated • 8 -
parameterlab/apricot_clustering_coqa_deberta-v3-base_for_vicuna-7b-v1.5
Text Classification • 0.2B • Updated • 10