The Alignment Problem: From RLHF and PPO to DPO and KTO
AlignmentSupervised Fine-tuning
👋 Welcome to my Blog
Hi, this is Chengshuo. I'm documenting my learning notes in this blog. Feel free to email me if there is any mistakes in the notes. 😉
Bypassing the Reward Model: Direct Preference Optimization (DPO)
LLM FundamentalsAlignment
放弃RLHF，拥抱DPO？
Alignment
The Alchemy of Weights: A Guide to Model Merging Techniques
Model ArchitectureFine-Tuning
Scaling Efficiently: Understanding Mixture of Experts (MoE)
Model ArchitectureScaling Laws
Stretching the Horizon: Extending LLM Context Windows with RoPE
Model ArchitectureLLM Fundamentals
Speeding Up the Brain: The Power of Semantic Caching for LLMs
System DesignPerformance Optimization
Breaking the Memory Bandwidth Bottleneck: A Deep Dive into Speculative Decoding
LLM InferencePerformance Optimization
Breaking the Memory Bottleneck: PagedAttention and RadixAttention
Inference OptimizationLLM Fundamentals
Categories
- All (53)
- AI Engineering (3)
- AI Safety (2)
- AI Search (3)
- Agent (9)
- Agentic RAG (1)
- Alignment (6)
- DNS (1)
- Embedding & Vector Database (3)
- Fine-Tuning (2)
- GraphRAG (1)
- Inference Optimization (4)
- Knowledge Management (1)
- LLM Evaluation (3)
- LLM Fundamentals (11)
- LLM Inference (1)
- Learning Systems (1)
- Model Architecture (4)
- Performance Optimization (3)
- Personal Reflextion (1)
- Post-Training (1)
- Programming Languages (1)
- Prompt Engineering (6)
- RAG (4)
- Scaling Laws (1)
- Search (1)
- Supervised Fine-tuning (5)
- System Design (3)
- Vibe Coding (1)
- Web Deployment (1)