Please enable JavaScript.
Coggle requires JavaScript to display documents.
LLM - Coggle Diagram
LLM
Building Blocks
Attention
MHA
MQA
GQA
MLA
Positional Embeddings
RoPE
Transformer
The original transformer architecture encoder-decoder
Activations
SwiGLU
GLU
GELU
LayerNorm
RMSNorm
State Space models
Diffusion LLMs
Inference
Speculative Decoding
Paged Attention
RAG
Embeddings Models
Training
Vector Databased
Re-ranker
KV Cache
Pre-fill
Introduction
Training
Evaluations
Data
Infrastructure