SRA achieves 99.71% average attack success across 26 LLMs by optimizing for coherent malicious semantics via the SRHS algorithm, with claimed theoretical guarantees on convergence and transfer.
Vicuna: An open-source chatbot impressing gpt- 4 with 90%* chatgpt quality
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
baseline 1
other 1
citation-polarity summary
verdicts
UNVERDICTED 4representative citing papers
PivotMerge merges heterogeneous multimodal pre-trained models via shared-space decomposition to filter conflicts and layer-wise weights based on alignment contributions, outperforming baselines on multimodal benchmarks.
ForgeryGPT integrates a forgery localization expert and mask encoder into an LLM for pixel-level forgery detection, localization, and explainable output via three-stage training on custom mask-text and instruction datasets.
Paraesthesia is an emotion-style dynamic backdoor attack achieving ~99% success rate on instruction and classification tasks across four LLMs while preserving clean performance.
citing papers explorer
-
LLM-Agnostic Semantic Representation Attack
SRA achieves 99.71% average attack success across 26 LLMs by optimizing for coherent malicious semantics via the SRHS algorithm, with claimed theoretical guarantees on convergence and transfer.
-
PivotMerge: Bridging Heterogeneous Multimodal Pre-training via Post-Alignment Model Merging
PivotMerge merges heterogeneous multimodal pre-trained models via shared-space decomposition to filter conflicts and layer-wise weights based on alignment contributions, outperforming baselines on multimodal benchmarks.
-
ForgeryGPT: A Multimodal LLM for Interpretable Image Forgery Detection and Localization
ForgeryGPT integrates a forgery localization expert and mask encoder into an LLM for pixel-level forgery detection, localization, and explainable output via three-stage training on custom mask-text and instruction datasets.
-
When Emotion Becomes Trigger: Emotion-style dynamic Backdoor Attack Parasitising Large Language Models
Paraesthesia is an emotion-style dynamic backdoor attack achieving ~99% success rate on instruction and classification tasks across four LLMs while preserving clean performance.