DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMs
【DL輪読会】DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantize more
Author: riyo.ono
1 RomanSetu: Efficiently unlocking multilingual capabilities of Large Language Models via Romanization (ACL 2024) + An Empirical Comparison of Vocabulary Expansion and Initialization Approaches for Language Models (2024)
【DL輪読会】RomanSetu: Efficiently unlocking multilingual capabilities of Large Language M more
Denoising Diffusion Bridge Models
【DL輪読会】Denoising Diffusion Bridge Models by @DeepLearning2023
Classifier-Free Guidance is a Predictor-Corrector
【DL輪読会】Classifier-Free Guidance is a Predictor-Corrector by @DeepLearning2023
DiG: Scalable and Efficient Diffusion Models with Gated Linear Attention
<script async class=”docswell-embed” src=”https://bcdn.docswell. more