Notes

02 elasticsearch

Initializing search

Blog Notes
01 agents system design
02 jax
02 llm modeling
02 llm models popular
02 open source tools
02 python3
02 pytorch

Notes

Blog Notes
01 agents system design
01 agents system design
02 jax
02 jax
- JAX: Composable Transformations for Numerical Computing
- 01 fundamentals
  01 fundamentals
- 02 core transformations
  02 core transformations
- 03 advanced patterns
  03 advanced patterns
- 04 machine learning
  04 machine learning
- 05 production & performance
  05 production & performance
02 llm modeling
02 llm modeling
- LLM Optimization & Modeling Knowledge Base
- 00 fundamentals
  00 fundamentals
- 01 architecture
  01 architecture
  - 01-Architecture: LLM Design & Optimization
  - 01 core designs
    01 core designs
    
    Mixture of Experts (MoE): Scaling to Trillions of Parameters
    
    Multi-Query Attention (MQA) & Grouped Query Attention (GQA)
    
    Rotary Position Embeddings (RoPE)
    
    Sliding Window Attention: Efficient Long-Context Attention
  - 02 attention optimization
    02 attention optimization
    
    Flash Attention v2: Improvements and Differences from v1
    
    Flash Attention: Complete Technical Guide
    
    Kernel Fusion: Complete Technical Guide
  - 03 memory management
    03 memory management
    
    KV Cache: Comprehensive Technical Guide
    
    PagedAttention: Detailed Technical Guide
  - 04 batching
    04 batching
    
    Continuous Batching: Complete Technical Guide
    
    Medusa: Multi-Head Decoding for Faster Inference
    
    Speculative Decoding: Complete Technical Guide
- 02 training
  02 training
  - 01 efficiency
    01 efficiency
    
    Distributed Training: Multi-GPU and Multi-Node LLM Training
    
    Gradient Checkpointing: Trading Compute for Memory
    
    Mixed Precision Training (AMP): Speed and Memory with Reduced Precision
  - 02 scaling
    02 scaling
    
    Scaling Laws & Optimal Allocation: Computing the Perfect Model Size
  - 03 compression
    03 compression
    
    Model Distillation: Compressing Knowledge into Smaller Models
    
    Pruning & Sparsity: Removing Redundant Parameters and Activations
    
    Token Merging: Reducing Attention Compute by Merging Tokens
  - 04 adaptation
    04 adaptation
    
    Adapter Methods Beyond LoRA: Efficient Fine-tuning Alternatives
    
    LoRA: Low-Rank Adaptation - Complete Technical Guide
    
    QLoRA: Quantized LoRA - Complete Comparison with LoRA
  - 05 alignment
    05 alignment
    
    Rlhf
  - Finetuning
    Finetuning
    
    Domain-Specific Fine-tuning: Adapting to Your Industry
    
    DPO: Direct Preference Optimization - Learning from Preferences Without RL
    
    Fine-tuning Best Practices: Tips, Tricks & Common Pitfalls
    
    Fine-tuning Fundamentals: Adapting Models to Your Task
    
    Instruction Tuning & SFT: Teaching Models to Follow Instructions
- 03 inference
  03 inference
  - 01 optimization
    01 optimization
    
    Decoding Strategies: Beam Search vs Greedy vs Sampling
    
    Length Extrapolation & Long Context: Training for Generalization to Longer Sequences
    
    LLM Inference Optimization: Complete Guide to Production Serving
  - 02 quantization
    02 quantization
    
    AWQ Quantization: Complete Technical Guide
    
    GPTQ Quantization: Complete Technical Guide
  - 03 knowledge integration
    03 knowledge integration
    
    Rag
    Rag
    
    Chunking & Document Preparation: Preparing Documents for RAG
    
    RAG Evaluation & Metrics: Measuring Performance
    
    Retrieval-Augmented Generation (RAG): Fundamentals
    
    RAG Integration & Prompt Engineering: Using Retrieved Context Effectively
    
    Reranking & Ranking: Improving Retrieval Quality
    
    Retrieval Strategies: Finding the Right Documents
    
    Vector Databases & Embeddings: The Foundation of RAG
  - Frameworks
    Frameworks
    
    Framework Comparison & Selection Guide
    
    Inference Frameworks Fundamentals: Choosing the Right Tool
    
    vLLM: Fast and Easy LLM Inference
- 04 production
  04 production
  - 01 deployment
    01 deployment
    
    Agent Framework Integration: LLM Deployment in Agentic Systems
    
    Cost Optimization Strategies: Reducing Training and Inference Expenses
    
    Load Balancing & Request Routing: Efficient Multi-Model Serving
  - 02 operations
    02 operations
    
    Model Merging & Ensemble Methods: Combining Models for Better Performance
    
    Monitoring & Observability: Production LLM Systems
    
    Safety & alignment
    Safety & alignment
    
    Prompt Injection & Adversarial Examples: Attack Patterns & Defenses
    
    Safety & Alignment Fundamentals: Keeping LLMs Safe and Aligned
- 05 benchmarks
  05 benchmarks
02 llm models popular
02 llm models popular
02 open source tools
02 open source tools
- Open-Source Tools for LLM Development
- 01 training
  01 training
- 02 inference
  02 inference
- 03 agent building
  03 agent building
- 04 knowledge management
  04 knowledge management
- 05 tool integration
  05 tool integration
  - Part 5: Tool Integration & Orchestration
  - Composio: Universal Tool Integration
- 06 monitoring
  06 monitoring
- 07 infrastructure
  07 infrastructure
02 python3
02 python3
- Python for ML & LLM Inference
- How Python Features Enable ML Frameworks
- Complete Outline: Python for ML & LLM Inference
- Python for ML & LLM Inference - Quick Start
- 01 fundamentals
  01 fundamentals
- 02 object oriented patterns
  02 object oriented patterns
- 03 functional programming
  03 functional programming
- 04 c extensions & ffi
  04 c extensions & ffi
- 05 memory & performance
  05 memory & performance
- 05 memory\ \&\ performance
  05 memory\ \&\ performance
- 06 dynamic features
  06 dynamic features
- 07 advanced typing
  07 advanced typing
- 08 module system
  08 module system
- 09 bytecode & execution
  09 bytecode & execution
- 10 ml specific patterns
  10 ml specific patterns
02 pytorch
02 pytorch
- 🧠 Advanced PyTorch Techniques
- 01 foundations & tensor mastery
  01 foundations & tensor mastery
- 02 module & layer engineering
  02 module & layer engineering
- 03 custom autograd & model transformation
  03 custom autograd & model transformation
- 04 performance & compilation
  04 performance & compilation
- 05 distributed training
  05 distributed training
- 06 mixed precision & quantization
  06 mixed precision & quantization
- 07 export, deployment & production
  07 export, deployment & production
- 08 advanced training techniques
  08 advanced training techniques
- 09 custom c++ and cuda extensions
  09 custom c++ and cuda extensions

02 elasticsearch

Copyright © 2024-2026 Jayaprakash

Made with Material for MkDocs