VFlash: Fast, Accurate Visual Planning via Flash Primitive Decoding
u_Unrolled
Easy10 XPattention

Why do GPT models use a causal mask?

What does the causal attention mask in a decoder-only model prevent?

Stuck? Show the solution

Answer: B. A token attending to tokens after it, which would leak the answer during training

Training predicts every next token in parallel. Without the mask, position t could see token t+1, the very thing it must predict.

Learn the idea behind itLLM Architecture Families — Trade-offs, Not Model Names