Skip to content

Add Support for GPT-2 Training on different Devices #551

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ShawnXuan merged 11 commits into from
Sep 13, 2024
Merged

Add Support for GPT-2 Training on different Devices #551

Changes from 1 commit
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
d5bdefe
fix bfloat16
ShawnXuan Sep 9, 2024
526f280
update tokenizer encode
ShawnXuan Sep 9, 2024
6ee4729
support python 3.10-
ShawnXuan Sep 9, 2024
f1960ca
support python 3.10-
ShawnXuan Sep 9, 2024
bef17f2
support python 3.10-
ShawnXuan Sep 9, 2024
c55ac20
click
ShawnXuan Sep 9, 2024
9c60405
fix tokenizer path
ShawnXuan Sep 9, 2024
b73104f
update
ShawnXuan Sep 10, 2024
12843f1
update
ShawnXuan Sep 10, 2024
c1d315e
checkout main files
ShawnXuan Sep 10, 2024
34052ed
Merge branch 'main' of github.com:Oneflow-Inc/libai into gpt_devices
ShawnXuan Sep 10, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
update
  • Loading branch information
@ShawnXuan
ShawnXuan committed Sep 10, 2024
commit b73104f2e098902eed595a3104e14b57f941afa8
4 changes: 3 additions & 1 deletion libai/models/gpt_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -244,7 +244,9 @@ def forward(self, input_ids, past_length=0):
bsz, seq_length = input_ids.size()

position_ids = self.position_ids[:, past_length : past_length + seq_length]
position_ids = position_ids.expand_as(input_ids).to_global(sbp=input_ids.sbp)
position_ids = position_ids.expand_as(input_ids).to_global(
sbp=input_ids.sbp, placement=input_ids.placement
)

token_embeds = self.token_embeddings(input_ids)
position_embeds = self.position_embeddings(position_ids)
Expand Down