Why is there no 512

#20
by Pluche - opened

Why is there no 512?
image.png

image.png

512 is the Sequence Length, whereas the numbers in 2_Dense_... refer to the dimensionality, i.e. the number of values that make up the text embeddings.
All of the different dimensionalities use a sequence length of 512 tokens.

  • Tom Aarsen
Pluche changed discussion status to closed

But 512 is mentioned in The models are finally trained by MRL, so they have multiple dimensions: 512, 768, 1024, 2048, 4096, 6144 and 8192.
256 is not mentioned but present in the repository. And it makes sense to have 512 between 256 and 768.

Your need to confirm your account before you can post a new comment.

Sign up or log in to comment