Skip to content

Latest commit

 

History

History
68 lines (37 loc) · 1.15 KB

File metadata and controls

68 lines (37 loc) · 1.15 KB

🤖Hand-Drawn-LLM🤖

Visualizing LLM Architecture Details using Excalidraw🎨🎨🎨

IF you like this repo, remember to follow. Your stars are also the motivation behind my updates.🤗🤗🤗

Family Model State
Qwen Qwen3.5
Qwen2.5_VL
Meta llama-2
OPENAI GPT-2
CLIP
GOOGLE ViT
GLM - update

Qwen3.5

Qwen3.5 Model

Vision Encoder

Text Encoder

Qwen-VL

CLIP

ViT

Qwen

✍️Qwen's architecture is similar to that of Llama; the key difference lies in the attention computation process.

Llama

GPT-2

Attention