mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
1a55fb0ee8
This PR is part of a series attempting to re-submit https://github.com/pytorch/pytorch/pull/134592 as smaller PRs. In jit tests: - Add and use a common raise_on_run_directly method for when a user runs a test file directly which should not be run this way. Print the file which the user should have run. - Raise a RuntimeError on tests which have been disabled (not run) Pull Request resolved: https://github.com/pytorch/pytorch/pull/154725 Approved by: https://github.com/Skylion007
129 lines
3.8 KiB
Python
129 lines
3.8 KiB
Python
# Owner(s): ["oncall: jit"]
|
|
|
|
import io
|
|
import unittest
|
|
|
|
import torch
|
|
from torch.testing._internal.common_utils import (
|
|
IS_WINDOWS,
|
|
raise_on_run_directly,
|
|
TEST_MKL,
|
|
)
|
|
from torch.testing._internal.jit_utils import JitTestCase
|
|
|
|
|
|
class TestSparse(JitTestCase):
|
|
def test_freeze_sparse_coo(self):
|
|
class SparseTensorModule(torch.nn.Module):
|
|
def __init__(self) -> None:
|
|
super().__init__()
|
|
self.a = torch.rand(3, 4).to_sparse()
|
|
self.b = torch.rand(3, 4).to_sparse()
|
|
|
|
def forward(self, x):
|
|
return x + self.a + self.b
|
|
|
|
x = torch.rand(3, 4).to_sparse()
|
|
|
|
m = SparseTensorModule()
|
|
unfrozen_result = m.forward(x)
|
|
|
|
m.eval()
|
|
frozen = torch.jit.freeze(torch.jit.script(m))
|
|
|
|
frozen_result = frozen.forward(x)
|
|
|
|
self.assertEqual(unfrozen_result, frozen_result)
|
|
|
|
buffer = io.BytesIO()
|
|
torch.jit.save(frozen, buffer)
|
|
buffer.seek(0)
|
|
loaded_model = torch.jit.load(buffer)
|
|
|
|
loaded_result = loaded_model.forward(x)
|
|
|
|
self.assertEqual(unfrozen_result, loaded_result)
|
|
|
|
def test_serialize_sparse_coo(self):
|
|
class SparseTensorModule(torch.nn.Module):
|
|
def __init__(self) -> None:
|
|
super().__init__()
|
|
self.a = torch.rand(3, 4).to_sparse()
|
|
self.b = torch.rand(3, 4).to_sparse()
|
|
|
|
def forward(self, x):
|
|
return x + self.a + self.b
|
|
|
|
x = torch.rand(3, 4).to_sparse()
|
|
m = SparseTensorModule()
|
|
expected_result = m.forward(x)
|
|
|
|
buffer = io.BytesIO()
|
|
torch.jit.save(torch.jit.script(m), buffer)
|
|
buffer.seek(0)
|
|
loaded_model = torch.jit.load(buffer)
|
|
|
|
loaded_result = loaded_model.forward(x)
|
|
|
|
self.assertEqual(expected_result, loaded_result)
|
|
|
|
@unittest.skipIf(IS_WINDOWS or not TEST_MKL, "Need MKL to run CSR matmul")
|
|
def test_freeze_sparse_csr(self):
|
|
class SparseTensorModule(torch.nn.Module):
|
|
def __init__(self) -> None:
|
|
super().__init__()
|
|
self.a = torch.rand(4, 4).to_sparse_csr()
|
|
self.b = torch.rand(4, 4).to_sparse_csr()
|
|
|
|
def forward(self, x):
|
|
return x.matmul(self.a).matmul(self.b)
|
|
|
|
x = torch.rand(4, 4).to_sparse_csr()
|
|
|
|
m = SparseTensorModule()
|
|
unfrozen_result = m.forward(x)
|
|
|
|
m.eval()
|
|
frozen = torch.jit.freeze(torch.jit.script(m))
|
|
|
|
frozen_result = frozen.forward(x)
|
|
|
|
self.assertEqual(unfrozen_result.to_dense(), frozen_result.to_dense())
|
|
|
|
buffer = io.BytesIO()
|
|
torch.jit.save(frozen, buffer)
|
|
buffer.seek(0)
|
|
loaded_model = torch.jit.load(buffer)
|
|
|
|
loaded_result = loaded_model.forward(x)
|
|
|
|
self.assertEqual(unfrozen_result.to_dense(), loaded_result.to_dense())
|
|
|
|
@unittest.skipIf(IS_WINDOWS or not TEST_MKL, "Need MKL to run CSR matmul")
|
|
def test_serialize_sparse_csr(self):
|
|
class SparseTensorModule(torch.nn.Module):
|
|
def __init__(self) -> None:
|
|
super().__init__()
|
|
self.a = torch.rand(4, 4).to_sparse_csr()
|
|
self.b = torch.rand(4, 4).to_sparse_csr()
|
|
|
|
def forward(self, x):
|
|
return x.matmul(self.a).matmul(self.b)
|
|
|
|
x = torch.rand(4, 4).to_sparse_csr()
|
|
m = SparseTensorModule()
|
|
expected_result = m.forward(x)
|
|
|
|
buffer = io.BytesIO()
|
|
torch.jit.save(torch.jit.script(m), buffer)
|
|
buffer.seek(0)
|
|
loaded_model = torch.jit.load(buffer)
|
|
|
|
loaded_result = loaded_model.forward(x)
|
|
|
|
self.assertEqual(expected_result.to_dense(), loaded_result.to_dense())
|
|
|
|
|
|
if __name__ == "__main__":
|
|
raise_on_run_directly("test/test_jit.py")
|