qwen35: use post-norm hidden state for MTP #24025
am17an
marked this pull request as draft 16 days ago
qwen35: use post-norm hidden state for MTP
e94341d3
rename pre_norm to nextn
7e00869e
fix step35
13d07d25
am17an
force pushed
from
9140cb9d
to
13d07d25
15 days ago
am17an
marked this pull request as ready for review 15 days ago
am17an
requested a review
15 days ago
am17an
requested a review
15 days ago
ggerganov
approved these changes
on 2026-06-03
am17an
merged
166fe294
into master 15 days ago
Assignees
No one assigned
Labels
model
examples
server
Login to write a write a comment.
Login via GitHub