Q, K and V
EXDai
• Datasets:
Tasks:
Feature Extraction
Modalities:
Image
Formats:
imagefolder
Languages:
English
Size:
< 1K
License:
Search is not available for this dataset
image imagewidth (px) 1.18k 2.38k |
|---|
Q, K and V — Inside a Transformer's Attention Projection
Companion dataset for Episode 12 of EXD: Q, K and V projection.
One input vector, three weight matrices, three purposes. This notebook loads Qwen3.6-35B-A3B and extracts the actual Q, K, V projections from a real attention layer.
📖 Article: huggingface.co/blog/EXDai/qkv
Run the notebook: download qkv_projection.ipynb and run it with a GPU.
- Downloads last month
- 251