Grokking Generative AI Fundamentals
0% completed
How Text Gets Generated
A customer asks a support assistant, "Where is my refund?" The table below follows the reply as it appears, one token at a time. The times are illustrative.
| Time | What happens |
|---|---|
| 0 ms | The request reaches the model. |
| 350 ms | "Your" |
| 380 ms | "refund" |
| 410 ms | "has" |
| 440 ms | "been" |
| 470 ms | "approved" |
| 500 ms | "." |
| 530 ms | A special token that means "done" |
The first token takes the longest because the model reads the whole request first
.....
.....
.....
Like the course? Get enrolled and start learning!
Reading Progress
0%