Zhongzhu's Blog

04-16

PipeDream: Turning Pipeline Parallelism into a Practical Training System — Deep Technical Review

04-16

PipeDream：把 Pipeline Parallelism 做成真正可训练系统——深度技术评审

04-15

LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding — Deep Technical Review

04-15

LayerSkip：让大模型“提前退出 + 自校验推理”成为可部署方案——深度技术评审

04-14

Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts — Deep Technical Review

04-14

ArmoRM：用“多目标奖励建模 + 混合专家门控”做可解释偏好学习——深度技术评审

04-13

Toolformer: Language Models Can Teach Themselves to Use Tools — Deep Technical Review

04-13

Toolformer：让语言模型自己学会“什么时候调用工具”——深度技术评审

04-12

Voyager: An Open-Ended Embodied Agent with Large Language Models — Deep Technical Review

04-12

Voyager：一个能在 Minecraft 中持续成长的 LLM 具身智能体 —— 深度技术评审