transformers
DeepSpeed: hardcode `torch.arange` dtype on `float` usage to avoid incorrect initialization
#28760

Merged

DeepSpeed: hardcode `torch.arange` dtype on `float` usage to avoid incorrect initialization #28760

gante merged 5 commits into huggingface:main from gante:deepspeed_init_fix

hardcode torch.arange type

ed5afa16

derp

408d8a3a

derp

f6756bcd

gante marked this pull request as ready for review 2 years ago

gante requested a review from

ArthurZucker 2 years ago

ArthurZucker commented on 2024-01-30

add test

b040282e

test comments

bd5d28d4

ArthurZucker approved these changes on 2024-01-31

gante merged beb2a096 into main 2 years ago

gante deleted the deepspeed_init_fix branch 2 years ago

Reviewers

ArthurZucker

Assignees

No one assigned

Labels

None yet

Milestone

No milestone