Skip to content

fix: index_put converter to handle multi-shape slicing with None #3465

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

Closed
wants to merge 1 commit into from
Closed

fix: index_put converter to handle multi-shape slicing with None #3465

Show file tree
Hide file tree
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
117 changes: 73 additions & 44 deletions py/torch_tensorrt/dynamo/conversion/impl/select.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -501,7 +501,6 @@ def index_put_converter(
F = [i for i in range(rank) if indices[i] is None] # Free dimensions
I = [i for i in range(rank) if indices[i] is not None] # Indexed dimensions
K = len(I)

# Determine the maximum size 'N' among the index tensors
if K > 0:
index_shapes = [tensor.shape[0] for tensor in indices if tensor is not None]
Expand Down Expand Up @@ -684,16 +683,6 @@ def index_put_converter(
values_reshaped = impl.shuffle.reshape(
ctx, target, source_ir, f"{name}_reshape_scalar", values, (1,)
)
num_dims = len(expected_shape)
ones_shape = tuple([1] * num_dims)
values_reshaped = impl.shuffle.reshape(
ctx,
target,
source_ir,
f"{name}_reshape_to_ones",
values_reshaped,
ones_shape,
)
values_expanded = impl.slice.expand(
ctx,
target,
Expand All @@ -704,40 +693,79 @@ def index_put_converter(
)
else: # Non-scalar case
values_shape = list(values.shape)

# Pad dimensions if necessary
if len(values_shape) < len(expected_shape):
values_shape = [1] * (
len(expected_shape) - len(values_shape)
) + values_shape

# Calculate a broadcastable shape
broadcast_shape = []
for exp_dim, val_dim in zip(expected_shape, values_shape):
if val_dim == 1:
broadcast_shape.append(exp_dim)
elif val_dim == exp_dim:
broadcast_shape.append(val_dim)
if K > 0 and N in values_shape:
n_idx = values_shape.index(N)
permute_order = [n_idx] + [
i for i in range(len(values_shape)) if i != n_idx
]
values_permuted = impl.permutation.permute(
ctx, target, source_ir, f"{name}_permute_values", values, permute_order
)
remaining_shape = [
values_shape[i] for i in range(len(values_shape)) if i != n_idx
]
target_f_dims = len(F)
current_f_dims = len(remaining_shape)
if current_f_dims < target_f_dims:
values_expanded_shape = (
[N] + [1] * (target_f_dims - current_f_dims) + remaining_shape
)
else:
raise ValueError(f"Cannot broadcast {values_shape} to {expected_shape}")

# Reshape and then expand
values_reshaped = impl.shuffle.reshape(
ctx,
target,
source_ir,
f"{name}_reshape_values",
values,
tuple(broadcast_shape),
)
values_expanded = impl.slice.expand(
ctx,
target,
source_ir,
f"{name}_expand_values",
values_reshaped,
expected_shape,
)
values_expanded_shape = [N] + remaining_shape[:target_f_dims]
values_expanded = impl.shuffle.reshape(
ctx,
target,
source_ir,
f"{name}_unsqueeze_values",
values_permuted,
tuple(values_expanded_shape),
)
broadcast_shape = []
for exp_dim, val_dim in zip(expected_shape, values_expanded_shape):
if val_dim == 1:
broadcast_shape.append(exp_dim)
elif val_dim == exp_dim:
broadcast_shape.append(val_dim)
else:
raise ValueError(
f"Cannot broadcast {values_expanded_shape} to {expected_shape}"
)
values_expanded = impl.slice.expand(
ctx,
target,
source_ir,
f"{name}_expand_values",
values_expanded,
tuple(broadcast_shape),
)
else:
values_shape_padded = [1] * (
len(expected_shape) - len(values.shape)
) + list(values.shape)
broadcast_shape = []
for exp_dim, val_dim in zip(expected_shape, values_shape_padded):
if val_dim == 1 or exp_dim == val_dim:
broadcast_shape.append(exp_dim)
else:
raise ValueError(
f"Cannot broadcast {values.shape} to {expected_shape}"
)
values_reshaped = impl.shuffle.reshape(
ctx,
target,
source_ir,
f"{name}_reshape_values",
values,
tuple(broadcast_shape),
)
values_expanded = impl.slice.expand(
ctx,
target,
source_ir,
f"{name}_expand_values",
values_reshaped,
expected_shape,
)

# Flatten values to (N * F_volume,)
flattened_values = impl.shuffle.reshape(
Expand All @@ -749,6 +777,7 @@ def index_put_converter(
(N * F_volume,),
)

indices_cat = cast_trt_tensor(ctx, indices_cat, trt.int32, f"{name}_idx_int32")
# Perform Scatter ND operation
scatter_layer = ctx.net.add_scatter(
input_tensor,
Expand Down
6 changes: 6 additions & 0 deletions tests/py/dynamo/conversion/test_index_put_aten.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,6 +194,12 @@ class TestIndexPutConverter(DispatchTestCase):
dtype=torch.int32,
),
),
param(
test_name="4d_indices_none_none_multiple_idx_broadcast_error",
source_tensor=torch.zeros([1, 2, 5, 3], dtype=torch.float32),
indices_tensor=(None, None, torch.tensor([0, 1, 2], dtype=torch.int64)),
value_tensor=torch.randn([2, 3, 3], dtype=torch.float32),
),
# param(
# test_name="2d_indices_accumulate_True",
# source_tensor=torch.zeros([5, 5], dtype=torch.int32),
Expand Down
Loading