eqho
/
LLM Lab
model
lab-tiny
the run
06 / 09
Lesson 06
Predicting the next token
The output is not a word. It is a probability for every word.
Next: Training
01
How big is a model?
02
Text becomes tokens
03
Tokens become vectors
04
Attention
05
The transformer block
06
Predicting the next token
Logits
Softmax and temperature
Top-k, top-p, sample
Feed it back in
07
Training
08
From autocomplete to assistant
09
Inference in production
01 / 01
Logits
One score per vocabulary entry. Bigger is more likely, but they are not probabilities yet.
←
→
space