来自 @爱可可-爱生活 的第一手AI快报，用最通俗的语言，聊最前沿的人工智能科研进展～ #人工智能# #科技前沿#
[CV] Align Your Flow: Scaling Continuous-Time Flow
[LG] Taming Polysemanticity in LLMs: Provable Feat
[LG] Flat Channels to Infinity in Neural Loss Land
[LG] GrokAlign: Geometric Characterisation and Acc
真正的自由，从来不是拥有无限的选择权。
[CL] AceReason-Nemotron 1.1: Advancing Math and Co
[LG] Random Matrix Theory for Deep Learning: Beyon
[LG] LiveCodeBench Pro: How Do Olympiad Medalists
[LG] Is your batch size the problem? Revisiting th
你会发现，决定你五年、十年后成为一个什么样的人，根本不是你那些写在纸上、存在手机里的宏伟计划，而是你
[LG] Because we have LLMs, we Can and Should Pursu
[LG] Text-to-LoRA: Instant Transformer Adaption[Sa
[LG] On the Closed-Form of Flow Matching: Generali
[CL] Don't Pay Attention[Avey AI]https://arxiv.org
真正的知识管理，不是收藏，而是创造。
[LG] AutoRule: Reasoning Chain-of-thought Extracte
[LG] HeurAgenix: Leveraging LLMs for Solving Compl
[LG] Dense SAE Latents Are Features, Not Bugs[MIT
[LG] Optimal Embedding Learning Rate in LLMs: The
[CL] Approximating Language Model Training Data fr
