Transformer Explainer · Dense GPT-2

TRANSFORMER EXPLAINER

make

Recorded GPT-2 example · first Generate downloads 626 MiBEnter generates; Shift+Enter adds a line break. Up to 32 tokens.

Temperature

0.8

Sampling

Top-k Top-p

k=5

Dense MoE MLA Local–global Looped Encoder-only Encoder–decoder Speech Mamba hybrid Linear attention Vision–language ViT CLIP SAM DiT Diffusion LM Protein Time series Training Speculative