03
1 篇
Zhang Haoyi
Hi, I'm Zhang Haoyi - HUST CS undergrad (2024). I'm an enthusiastic learner, developer and blogger. Welcome to my personal space where I share my thoughts, notes, projects, and experiences.
统计
74文章
8分类
223标签
目录
-
文章
-
ai_agent
- Agent Architecture —— LLM & Planning & Tool & Memory
- Agent Communication Protocols —— MCP & A2A
- Agent Engineering —— Prompt & Context & Harness & Loop
- Agent Paradigms —— ReAct & Plan-and-Solve & Reflection
- Function Calling —— 工具调用
- Memory Mechanisms Evolution: From Storage to Experience —— 智能体记忆机制的演化综述
- Memory System —— Agent记忆系统
- Multi-Agent System—— 多智能体系统
- Recursive Self-Improvement（RSI）—— 递归自我改进综述
- Retrieval-Augmented Generation —— 检索增强生成
- Self-Evolving Agents (What、When、How、Where) —— 自进化智能体综述
-
ai_alignment
- Comparison of RL Algorithm —— PPO、GRPO、DPO、DAPO、GSPO技术详解与对比
- DAPO (Decoupled Clip and Dynamic sAmpling Policy Optimization) -- 解耦裁剪与动态采样策略优化
- DPO (Direct Preference Optimization) -- 直接偏好优化
- GRPO (Group Relative Policy Optimization) -- 群体相对策略优化
- GSPO (Group Sequence Policy Optimization) -- 群组序列策略优化
- PEFT (Parameter-Efficient Fine-Tuning) —— 参数高效微调
- PPO(Proximal Policy Optimization) -- 近端策略优化
- Reinforcement Learning -- 策略梯度、优势函数、重要性采样与KL散度惩罚
- RLHF (Reinforcement Learning from Human Feedback) —— 基于人类反馈的强化学习
- RLVR (Reinforcement Learning with Verifiable Rewards) —— 可验证奖励强化学习
- SFT (Supervised Fine-Tuning)—— 监督微调
-
ai_multimodal
- Contrastive Language-Image Pre-training —— CLIP对比学习
- Denoising Diffusion Probabilistic Models —— DDPM扩散模型
- Multimodal Large Language Model —— MLLM多模态大语言模型
- Multimodal RAG —— 多模态RAG
- Video Audio Multimodal —— 视频理解 & 视频生成 & 音频处理
- Vision Transformer —— ViT视觉Transformer
- Vision-Language Model —— VLM视觉语言模型
- Vision-Language-Action —— VLA具身智能
-
cs336
-
deep_learning
- Activation & Initialization —— 激活函数与权重初始化
- Autograd & Computational Graph —— 自动微分与计算图
- CNN —— 卷积神经网络
- Gradient Descent Optimizer —— 梯度下降优化器
- LSTM & GRU —— 门控循环神经网络
- MLP & Back Propagation —— 多层感知机与反向传播
- Probability & Information —— 概率论与信息论基础
- Residual Network —— ResNet残差网络
- RNN & BPTT —— 循环神经网络与随时间反向传播
- Tensor Operations —— 向量、矩阵与张量运算
-
machine_learning
- Adaptive Boosting (AdaBoost) —— 自适应提升
- Bagging & Random Forest —— 随机森林
- Decision Tree —— 决策树
- Expectation-Maximization Algorithm —— EM算法
- Gradient Boosting Machine (GBM) —— 梯度提升机与加法模型
- Hidden Markov Model (HMM) —— 隐马尔可夫模型
- K-Means Clustering —— 聚类算法
- K-Nearest Neighbor (KNN) —— K-近邻
- Kernel Trick —— 核技巧与常用核函数
- Linear Regression —— 线性回归
- Logistic Regression —— 逻辑回归与Softmax多分类
- Naive Bayes —— 朴素贝叶斯
- Principal Component Analysis (PCA) —— 主成分分析
- Support Vector Machine (SVM) —— 支持向量机
- XGBoost & LightGBM —— 梯度提升框架
-
-
相册
-
IB_Course
-
Calculus
-
Data Strctures
-
Discrete Mathematics
-
Physics Experiment
-
Physics(1)
-
Probability and Statistic
-
-
IIA_Course
-
Algorithm
-
Digital Circuits
-
Physics(2)
-
-
IIB_Course
-
Computer Organization
-
Database
-
MaoZedong
-
Signal and Linear Systems
-
-
Scenery
-
-
项目展示
-
CME295
-
CS336
-
分类
标签
Zhang Haoyi
Hi, I'm Zhang Haoyi - HUST CS undergrad (2024). I'm an enthusiastic learner, developer and blogger. Welcome to my personal space where I share my thoughts, notes, projects, and experiences.
统计
74文章
8分类
223标签
TAG TIMELINE
#交叉熵损失
2 篇文章 · 标签时间线
2 篇文章
0120261 篇文章
0220251 篇文章