DeepSpeed
Fixed bug with hybrid engine generation when inference_tp_size > 1
#4493
Open

Fixed bug with hybrid engine generation when inference_tp_size > 1 #4493

hxdtest wants to merge 3 commits into deepspeedai:master from hxdtest:fix_attention_mask
hxdtest
gather attention mask
1a597655
hxdtest hxdtest requested a review from jeffra jeffra 1 year ago
hxdtest hxdtest requested a review from tjruwase tjruwase 1 year ago
hxdtest hxdtest changed the title gather attention mask Fixed bug with hybrid engine generation when inference_tp_size > 1 1 year ago
hxdtest Merge branch 'master' into fix_attention_mask
3c7dc490
tjruwase Merge branch 'master' into fix_attention_mask
e9eb917a
tjruwase tjruwase requested a review from lekurile lekurile 1 year ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone