Menu
Sign In Search Podcasts Charts People & Topics Add Podcast API Pricing
Podcast Image

AI可可AI生活

[人人能懂] 从养成通才、高效瘦身到精准遗忘

27 Nov 2025

Description

我们都知道AI越来越强大,但你有没有想过,我们该如何让它跑得更快、更稳,甚至更“多才多艺”?本期节目,我们将一起探索几篇最新论文,看看科学家们是如何给AI的训练过程装上一个更稳健的导航系统,并揭开AI绘画高手背后“民间偏方”的科学原理。我们还会聊到,如何像培养一个“通才”一样,让一个AI同时学会两百件事。最后,我们将见证两种神奇的“魔法”:如何在没有数据的情况下给大模型高效“瘦身”,以及如何对一个黑箱模型进行精准的“微创手术”。00:00:41 如何给AI装上一个更聪明的“导航系统”00:05:19 AI绘画高手,背后藏着什么训练秘诀?00:11:06 AI通才养成记:如何让一个机器学会200件事?00:17:12 AI模型“瘦身”,如何做到无米之炊?00:25:14 给AI模型做微创手术,需要几步?本期介绍的几篇论文:[LG] ROOT: Robust Orthogonalized Optimizer for Neural Network Training  [Huawei Noah’s Ark Lab]  https://arxiv.org/abs/2511.20626 ---[LG] Demystifying Diffusion Objectives: Reweighted Losses are Better Variational Bounds  [Google DeepMind]  https://arxiv.org/abs/2511.19664 ---[LG] Learning Massively Multitask World Models for Continuous Control  [University of California San Diego]  https://arxiv.org/abs/2511.19584 ---[LG] CafeQ: Calibration-free Quantization via Learned Transformations and Adaptive Rounding  [Google]  https://arxiv.org/abs/2511.19705 ---[LG] ModHiFi: Identifying High Fidelity predictive components for Model Modification  [CSA, IISc & HP Inc. AI Lab & Google]  https://arxiv.org/abs/2511.19566 

Audio
Featured in this Episode

No persons identified in this episode.

Transcription

This episode hasn't been transcribed yet

Help us prioritize this episode for transcription by upvoting it.

0 upvotes
🗳️ Sign in to Upvote

Popular episodes get transcribed faster

Comments

There are no comments yet.

Please log in to write the first comment.