很荣幸我们的论文《Learn from model Beyond Finetuning》被 nature machine intelligence 接收。这个项目在武大,中大,北理工,南洋理工多位老师的通力合作下完成,恭喜所有的合作者们。 这个工作传达的思想是:大模型时代模型与数据几乎是等价的,我们定义了 “从模型中学习” 的新范式,利用已有的开源模型以及训练checkpoint实现低成本高效率的AI模型制作。 你说的对,但是 FusionBench 是由研究人员们自主开发的一款开源的深度模型融合评测基准。基准测试发生在一个被称作「深度学习」的神奇领域,在这里,被精心挑选的模型将被赋予「融合之力」,整合多样化的特征表示。你将扮演一位名为「研究者」的角色,在多样化的数据集中探索各种融合策略,与不同的模型架构相遇,它们各自具备独特的性能特点。通过这些模型的协同合作,你将应对各种具有挑战性的任务,解决现实世界中的难题。
Exploring new planets, building advanced machinery, & conquering the cosmos…
A colossal spaceship equipped with huge science fiction-inspired antimatter engines is drifting through the vast, dark expanse of space. (prompt to generate the teaser image, using DALL-E 3) Plot Once upon a time, two fearless engineers crash-landed on a desolate, forsaken planet. With nothing but their wits and a few rusty tools, they began the…
In the ever-evolving landscape of machine learning, model fusion has emerged as a promising technique for enhancing model performance and generalization. Today, we’re excited to introduce FusionBench, a comprehensive benchmark specifically designed to evaluate and compare deep model fusion methodologies. FusionBench offers a robust, modular codebase that empowers researchers and developers to push the boundaries…
Are you tired of manually merging files and directories that have identical content? Look no further than merge_ln! This new script allows you to merge two files or two directories by creating hard links to identical files. Using the cmp and ln commands, merge_ln tests if two files are the same and merges them as…