Skip to content

Commit c816838

Browse files
committed
Follow up on merge conflict in release
1 parent b80029e commit c816838

File tree

1 file changed

+26
-25
lines changed
  • hyperpod-pytorch-job-template/hyperpod_pytorch_job_template/v1_1

1 file changed

+26
-25
lines changed

hyperpod-pytorch-job-template/hyperpod_pytorch_job_template/v1_1/model.py

Lines changed: 26 additions & 25 deletions
Original file line numberDiff line numberDiff line change
@@ -373,32 +373,33 @@ def build_dict(**kwargs):
373373
return {k: v for k, v in kwargs.items() if v is not None}
374374

375375
# Build resources
376-
if self.instance_type is None:
377-
requests_value = limits_value = {"nvidia.com/gpu": "0"}
376+
if self.accelerator_partition_type:
377+
partition_resource_key = f"nvidia.com/{self.accelerator_partition_type}"
378+
requests_value = build_dict(
379+
**{partition_resource_key: str(self.accelerator_partition_count)} if self.accelerator_partition_count else {},
380+
vcpu=str(self.vcpu) if self.vcpu else None,
381+
memory=str(self.memory) if self.memory else None,
382+
**{"vpc.amazonaws.com/efa": "1"} if self.instance_type and "p4d" in self.instance_type else {}
383+
)
384+
limits_value = build_dict(
385+
**{partition_resource_key: str(self.accelerator_partition_limit)} if self.accelerator_partition_limit else {},
386+
vcpu=str(self.vcpu_limit) if self.vcpu_limit else None,
387+
memory=str(self.memory_limit) if self.memory_limit else None,
388+
**{"vpc.amazonaws.com/efa": "1"} if self.instance_type and "p4d" in self.instance_type else {}
389+
)
378390
else:
379-
if self.accelerator_partition_type:
380-
partition_resource_key = f"nvidia.com/{self.accelerator_partition_type}"
381-
requests_value = build_dict(
382-
**{partition_resource_key: str(self.accelerator_partition_count)} if self.accelerator_partition_count else {},
383-
vcpu=str(self.vcpu) if self.vcpu else None,
384-
memory=str(self.memory) if self.memory else None
385-
)
386-
limits_value = build_dict(
387-
**{partition_resource_key: str(self.accelerator_partition_limit)} if self.accelerator_partition_limit else {},
388-
vcpu=str(self.vcpu_limit) if self.vcpu_limit else None,
389-
memory=str(self.memory_limit) if self.memory_limit else None
390-
)
391-
else:
392-
requests_value = build_dict(
393-
accelerators=str(self.accelerators) if self.accelerators else None,
394-
vcpu=str(self.vcpu) if self.vcpu else None,
395-
memory=str(self.memory) if self.memory else None
396-
)
397-
limits_value = build_dict(
398-
accelerators=str(self.accelerators_limit) if self.accelerators_limit else None,
399-
vcpu=str(self.vcpu_limit) if self.vcpu_limit else None,
400-
memory=str(self.memory_limit) if self.memory_limit else None
401-
)
391+
requests_value = build_dict(
392+
**{"nvidia.com/gpu": str(self.accelerators)} if self.accelerators else {},
393+
vcpu=str(self.vcpu) if self.vcpu else None,
394+
memory=str(self.memory) if self.memory else None,
395+
**{"vpc.amazonaws.com/efa": "1"} if self.instance_type and "p4d" in self.instance_type else {}
396+
)
397+
limits_value = build_dict(
398+
**{"nvidia.com/gpu": str(self.accelerators_limit)} if self.accelerators_limit else {},
399+
vcpu=str(self.vcpu_limit) if self.vcpu_limit else None,
400+
memory=str(self.memory_limit) if self.memory_limit else None,
401+
**{"vpc.amazonaws.com/efa": "1"} if self.instance_type and "p4d" in self.instance_type else {}
402+
)
402403

403404
# Build container
404405
container_kwargs = build_dict(

0 commit comments

Comments
 (0)