Skip to content

Skip caching_allocator_warmup on Neuron (no reuse pool to warm; currently OOMs) #47029

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
Cyrilvallez merged 5 commits into from
Jul 8, 2026
Merged

Skip caching_allocator_warmup on Neuron (no reuse pool to warm; currently OOMs) #47029

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
5987414
Skip caching_allocator_warmup on Neuron
dacorvo Jul 2, 2026
798bb59
fix: clean up comments for Neuron device handling
dacorvo Jul 3, 2026
b2bc0dd
Merge branch 'main' into skip_neuron_device_warmup
dacorvo Jul 7, 2026
860f8ab
Merge branch 'main' into skip_neuron_device_warmup
dacorvo Jul 7, 2026
2ff0c3f
fix comment
Cyrilvallez Jul 8, 2026
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
3 changes: 3 additions & 0 deletions src/transformers/modeling_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5135,6 +5135,9 @@ def caching_allocator_warmup(model: PreTrainedModel, expanded_device_map: dict,
# This causes the warmup function to return a `RuntimeError: Invalid buffer size: XX.XX GiB`.
# NOTE: not tested on intel macs
continue
elif device.type == "neuron":
# Skip warmup on Neuron (AWS Trainium/Inferentia): it provides no benefit as there is no reusable memory pool
continue
# We divide by 2 here as we allocate in fp16
_ = torch.empty(int(byte_count // 2), dtype=torch.float16, device=device, requires_grad=False)

Expand Down
Loading