riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-08-01 02:31:11 +06:00
Code Issues Actions 3 Packages Projects Releases Wiki Activity

Make ClipSeg compatible with model parallelism (#22844)

Browse Source
This commit is contained in:
Youssef Adarrab 2023-04-18 23:31:59 +00:00 committed by GitHub
parent 5bb4ec6233
commit 84a6570e7b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 2 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/transformers/models/clipseg/modeling_clipseg.py
View File

@ -1480,6 +1480,8 @@ class CLIPSegForImageSegmentation(CLIPSegPreTrainedModel):
loss = None
if labels is not None:
# move labels to the correct device to enable PP
labels = labels.to(logits.device)
loss_fn = nn.BCEWithLogitsLoss()
loss = loss_fn(logits, labels)