AI That Predicts Multiple Words at Once DeekSeek
Nothing is loaded from YouTube until you press play, and then without tracking cookies.
About this video
Unlock the secrets of multi-token prediction in transformer models! This video breaks down how transformers predict multiple tokens simultaneously, reusing expensive computations for efficiency. Learn how this approach mimics human thought processes and optimizes performance. #TransformerModels #MultiTokenPrediction #NLP #DeepLearning #AI #MachineLearning #NaturalLanguageProcessing #ArtificialIntelligence #ComputationalLinguistics #TransformerNetworks #nearai
Transcript
213 wordsCaptions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.
- 0:00
so so effectively we say well we we have the input token we did all the expensive computation of the Transformer and at the end we only predict one token but in many cases presumably you have enough information to predict a couple like if you think how we think we usually have like we don't need to know the next word we're about to say to know our thought right if it makes sense to reuse this expensive computation with the with a lot like on this picture something to
- 0:23
pay attention to is that in the leftmost block there's L Transformer blocks there's L layers and L is aett a high number it's like between 60 and 90 right uh but in the second and the third block here there's only one Transformer block right so what happens is that you feed uh the sequence of tokens like the first four tokens into the Transformer it goes through all the layers goes to the output layer predicts one token final
- 0:47
Vector that goes into the output layer it goes into one more Transformer block does that Transformer blocks goes into the output layer predicts the token after next right and so that gives you more signal that allows you to train