6 papers
Dynamic Shapley Computation
Xuan Yang, Hsi-Wen Chen, Ming-Syan Chen +1
Shapley-based data valuation provides a principled way to quantify the contribution of training data, but its high computational cost makes it impractical in dynamic settings where…
Interference-Aware Multi-Task Unlearning
Ying-Hua Huang, Rui Fang, Hsi-Wen Chen +1
Machine unlearning aims to remove the contribution of designated training data from a trained model while preserving performance on the remaining data. Existing work mainly focuses…
LoopQ: Quantization for Recursive Transformers
Rui Fang, Hsi-Wen Chen, Ming-Syan Chen
Looped language models (LoopLMs) improve parameter efficiency by recursively reusing Transformer blocks, enabling deeper computation under a fixed model size. However, this reuse m…
Amortized-Precision Quantization for Early-Exit Vision Transformers
Rui Fang, Hsi-Wen Chen, Ming-Syan Chen
Vision Transformers (ViTs) achieve strong performance across vision tasks, yet their deployment with low-precision early exiting remains fragile. Existing quantization methods assu…
Local Shapley: Model-Induced Locality and Optimal Reuse in Data Valuation
Xuan Yang, Hsi-Wen Chen, Ming-Syan Chen +1
The Shapley value provides a principled foundation for data valuation, but exact computation is #P-hard due to the exponential coalition space. Existing accelerations remain global…
Self-guided Knowledgeable Network of Thoughts: Amplifying Reasoning with Large Language Models
Chao-Chi Chen, Chin-Yuan Yeh, Hsi-Wen Chen +2
We introduce Knowledgeable Network of Thoughts (kNoT): a prompt scheme that advances the capabilities of large language models (LLMs) beyond existing paradigms like Chain-of-Though…