Prefill-related logic in input preparation for generation #42088
add prefill arg in generation
77f4b608
add a slow test
423a9cb8
fix copies
0f659bb5
can be like this but checking special tokens isn't good
906f88c6
ig this solves the issue with assisted_gen+prefill
5a918de8
update overwritten `prepare_inpits_for_generation`
ef04c518
prefill is actually when we have no cache at all.. Try this for now
00e4814e
first iteration is not always techincally same as prefill
6338d174
fix?
72916075
fix now?
32e54658
update bloom
375ad90e
fix smth
a184c8bb
make style
79abb96e
fix copies and skip test
36c60529
Merge branch 'main' into prefill-logic
1ff4e23e
Merge branch 'main' into prefill-logic
d4d99cb8
zucchini-nlp
changed the title [WIP] Prefill-related logic in input preparation for generation Prefill-related logic in input preparation for generation 209 days ago
fix copies
939e58dc
vasqu
commented
on 2025-11-19
tiny updates after a review
820cc920
fix other slow tests
597b1879
merge main
406bdb32
fix copies
39fb07ab
do not pass the same kwargs twice in prefill
cf1486c6
oops
1dc9cc2e
Merge branch 'main' into prefill-logic
2a44715c
have to revert? prob fails only on dgx
c82f5381
adjust slow test again
d156e10d
vasqu
approved these changes
on 2025-12-05
Merge branch 'main' into prefill-logic
b15723c2
address comments
028d2eb7
Merge branch 'main' into prefill-logic
a0a2ac7a
fix copies
ba023d0f
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub