Skip to content

test_ops: do not lower precision in conversion to Torch tensor #681

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
danieldk merged 3 commits into from
May 27, 2022
Merged

test_ops: do not lower precision in conversion to Torch tensor #681

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b55cc05
test_ops: do not lower precision in conversion to Torch tensor
danieldk May 27, 2022
400cc21
test_ops: compare arrays on same device in Torch comparison
danieldk May 27, 2022
88b17bb
test_maxout: compare arrays with same precision
danieldk May 27, 2022
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
5 changes: 1 addition & 4 deletions thinc/tests/backends/test_ops.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1259,16 +1259,13 @@ def test_ngrams():
def test_compare_activations_to_torch(ops, dtype, x, torch_func):
import torch

def cast_torch(scalar: float):
return torch.tensor([scalar], requires_grad=True)

func_name, pytorch_func = torch_func
forward = getattr(ops, func_name)
backward = getattr(ops, "backprop_" + func_name)
# The tolerance of isclose is set to 1e-06 instead of
# the default 1e-08 due to the GELU
x_thinc = ops.asarray([x], dtype=dtype)
x_torch = cast_torch(x)
x_torch = xp2torch(x_thinc, requires_grad=True)
y = pytorch_func(x_torch)
y_thinc = forward(x_thinc)
y.backward()
Expand Down