随缘随笔
Insights Flow

09-27

Synthetic, Paraphrase, Explain or Predict? 会比Pretrain+RL更优雅吗

12-14

重读STaR，与o1随想

05-27

论文阅读[精读]-Manyshot-ICL: 在context中重现传统AI的可能性

03-23

论文阅读[精读]-MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training

03-02

从DALL.E 3沿用到Sora的Recaption: GPT4也在用？和"Synthetic Data"是一个意思吗？

02-10

Self-Consistency之我见，兼More Agents is All You Need

12-19

论文阅读[粗读]-Alignment For Honesty

12-16

Weak-to-Strong Generalization(上): OpenAI是怎么看问题的？

11-28

论文阅读[精读]-Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamics

11-11

LLaVA, LLaVA 1.5和LLaVA-Plus: 讲讲LMM