Skip to content

Commit acf7fdd

Browse files
build: manually update PyTorch version (#4231)
This commit sets the PyTorch and TorchVision versions to nightly release 2025-06-15. This commit also disables a check for the test `test_broadcast_unit_dim_to_dynamic_with_unchanged_dim_dynamic` in `symbolic_shape_expr_test.py` because it generates different outputs for the torch stable and nightly versions. The check can be re-enabled once the nightly changes become part of the stable torch version. --------- Signed-off-by: Vivek Khandelwal <[email protected]>
1 parent 55c638e commit acf7fdd

File tree

4 files changed

+6
-4
lines changed

4 files changed

+6
-4
lines changed

pytorch-hash.txt

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
724da4eecc6c0ae561efbdcb371ac3f07a7ba8bd
1+
74d0136772a9564c77360c17c849c3a368fa3e8f

pytorch-requirements.txt

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
-f https://download.pytorch.org/whl/nightly/cpu/torch/
22
--pre
3-
torch==2.8.0.dev20250609
3+
torch==2.8.0.dev20250615

test/python/fx_importer/symbolic_shape_expr_test.py

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -371,7 +371,9 @@ def forward(self, x: torch.Tensor, y: torch.Tensor) -> torch.Tensor:
371371
# CHECK: torch.bind_symbolic_shape %[[ARG0]], [%[[S0]]], affine_map<()[s0] -> (1, s0)> : !torch.vtensor<[1,?],f32>
372372
# CHECK: torch.bind_symbolic_shape %[[ARG1]], [%[[S1]]], affine_map<()[s0] -> (s0)> : !torch.vtensor<[?],f32>
373373
# CHECK: %[[EXPAND:.+]] = torch.aten.expand %[[ARG0]], {{.*}}, {{.*}} : !torch.vtensor<[1,?],f32>, !torch.list<int>, !torch.bool -> !torch.vtensor<[?,?],f32>
374-
# CHECK: torch.bind_symbolic_shape %[[EXPAND]], [%[[S0]], %[[S1]]], affine_map<()[s0, s1] -> (s1, s0)> : !torch.vtensor<[?,?],f32>
374+
# TODO: Re-enable checks once the nightly changes are in stable.
375+
# The difference is that the nightly version generates a symbolic shape expression for the expand op with the order of dimensions swapped.
376+
# CHECK-DISABLED: torch.bind_symbolic_shape %[[EXPAND]], [%[[S1]], %[[S0]]], affine_map<()[s0, s1] -> (s0, s1)> : !torch.vtensor<[?,?],f32>
375377
# CHECK: return %[[EXPAND]] : !torch.vtensor<[?,?],f32>
376378
def test_broadcast_unit_dim_to_dynamic_with_unchanged_dim_dynamic():
377379
class BroadcastUnitDimToDynamicWithUnchangedDimDynamic(torch.nn.Module):

torchvision-requirements.txt

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
-f https://download.pytorch.org/whl/nightly/cpu/torchvision/
22
--pre
3-
torchvision==0.23.0.dev20250609
3+
torchvision==0.23.0.dev20250615

0 commit comments

Comments
 (0)