Skip to content

refactor: refactor test utils #1123

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

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

refactor: refactor test utils #1123

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
63 changes: 31 additions & 32 deletions tests/modeling_test_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -217,48 +217,47 @@ def set_dtype(model, dtype):
return model


def _generate_inputs(base_input, dtype, framework):
if isinstance(base_input, np.ndarray):
if base_input.dtype in (np.float16, np.float32, np.float64, bfloat16):
base_input = base_input.astype(dtype)

if framework == "torch":
return (
torch.from_numpy(base_input.astype(np.float32)).to(torch.bfloat16)
if dtype == bfloat16
else torch.from_numpy(base_input)
)
elif framework == "mindspore":
return ms.Tensor.from_numpy(base_input)
else:
raise ValueError(f"Unsupported framework: {framework}")

elif isinstance(base_input, (tuple, list)):
sequence_cls = type(base_input)
return sequence_cls(_generate_inputs(x, dtype, framework) for x in base_input)

elif isinstance(base_input, dict):
return {k: _generate_inputs(v, dtype, framework) for k, v in base_input.items()}

else:
return base_input


def generalized_parse_args(pt_dtype, ms_dtype, *args, **kwargs):
# parse args
pt_inputs_args = tuple()
ms_inputs_args = tuple()
for x in args:
if isinstance(x, np.ndarray):
if x.dtype in (np.float16, np.float32, np.float64, bfloat16):
px = x.astype(NP_DTYPE_MAPPING[pt_dtype])
mx = x.astype(NP_DTYPE_MAPPING[ms_dtype])
else:
px = mx = x

pt_inputs_args += (
(torch.from_numpy(px.astype(np.float32)).to(torch.bfloat16),)
if pt_dtype == "bf16"
else (torch.from_numpy(px),)
)
ms_inputs_args += (ms.Tensor.from_numpy(mx),)
else:
pt_inputs_args += (x,)
ms_inputs_args += (x,)
pt_inputs_args += (_generate_inputs(x, NP_DTYPE_MAPPING[pt_dtype], "torch"),)
ms_inputs_args += (_generate_inputs(x, NP_DTYPE_MAPPING[ms_dtype], "mindspore"),)

# parse kwargs
pt_inputs_kwargs = dict()
ms_inputs_kwargs = dict()
for k, v in kwargs.items():
if isinstance(v, np.ndarray):
if v.dtype in (np.float16, np.float32, np.float64, bfloat16):
px = v.astype(NP_DTYPE_MAPPING[pt_dtype])
mx = v.astype(NP_DTYPE_MAPPING[ms_dtype])
else:
px = mx = v

pt_inputs_kwargs[k] = (
torch.from_numpy(px.astype(np.float32)).to(torch.bfloat16)
if pt_dtype == "bf16"
else torch.from_numpy(px)
)
ms_inputs_kwargs[k] = ms.Tensor.from_numpy(mx)
else:
pt_inputs_kwargs[k] = v
ms_inputs_kwargs[k] = v
pt_inputs_kwargs[k] = _generate_inputs(v, NP_DTYPE_MAPPING[pt_dtype], "torch")
ms_inputs_kwargs[k] = _generate_inputs(v, NP_DTYPE_MAPPING[ms_dtype], "mindspore")

return pt_inputs_args, pt_inputs_kwargs, ms_inputs_args, ms_inputs_kwargs

Expand Down