/top/
/new/
/best/
/ask/
/show/
/job/
^
slacker news
login
about
←back to thread
Mercury: Ultra-fast language models based on diffusion
(arxiv.org)
566 points
PaulHoule
| 1 comments |
07 Jul 25 12:31 UTC
|
HN request time: 0.243s
|
source
1.
awaymazdacx5
◴[
07 Jul 25 14:49 UTC
]
No.
44490962
[source]
▶
>>44489690 (OP)
#
Having token embeddings with diffusion models, for 16x16 transformer encoding. Image is tokenized before transformers compile it. If decomposed virtualization modulates according to a diffusion model.
ID:
GO
↑