Skip to content

[configure_distributed_model] improve assert #1053

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
tjruwase merged 3 commits into from
May 13, 2021
Merged

[configure_distributed_model] improve assert #1053

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4a2716f
[configure_distributed_model] improve assert
stas00 May 7, 2021
b381e16
style
stas00 May 7, 2021
227c358
Merge branch 'master' into patch-1
tjruwase May 13, 2021
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
20 changes: 18 additions & 2 deletions deepspeed/runtime/engine.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -576,10 +576,26 @@ def _configure_distributed_model(self, model):
if self.zero_optimization_partition_weights() and any(
[hasattr(param,
'ds_id') for param in self.module.parameters()]):
assert all([param.dtype == torch.half for param in self.module.parameters()]), "fp16 is enabled but one or several model parameters have dtype that is not fp16"
if not all(
[param.dtype == torch.half for param in self.module.parameters()]):
names = [
n for n,
p in self.module.named_parameters() if p.dtype != torch.half
]
raise ValueError(
"fp16 is enabled but the following parameters have dtype that is not fp16: {', '.join(names)}"
)
self.module.half()
else:
assert all([param.dtype == torch.float for param in self.module.parameters()]), "fp16 is not enabled but one or several model parameters have dtype of fp16"
if not all(
[param.dtype == torch.float for param in self.module.parameters()]):
names = [
n for n,
p in self.module.named_parameters() if p.dtype != torch.float
]
raise ValueError(
"fp32 is enabled but the following parameters have dtype that is not fp32: {', '.join(names)}"
)

if not self.dont_change_device:
self.module.to(self.device)
Expand Down