Internal only change.

ai_edge_torch/generative/layers/experimental/xpu/__init__.py

@@ -0,0 +1,14 @@
+# Copyright 2025 The AI Edge Torch Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#

ai_edge_torch/generative/layers/experimental/xpu/split_attention_mask.py

@@ -0,0 +1,67 @@
+# Copyright 2025 The AI Edge Torch Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+"""Split attention mask."""
+
+
+import torch
+from torch import nn
+
+
+class SplitAttentionMask(nn.Module):
+  """Split attention mask."""
+
+  def __init__(self, context_size):
+    super().__init__()
+    self.context_size = context_size
+
+  def build_mask(self, num_input_tokens):
+    row_index, col_index = torch.meshgrid(
+        torch.arange(self.context_size),
+        torch.arange(self.context_size),
+        indexing='ij',
+    )
+    mask = torch.logical_or(
+        torch.logical_or(
+            torch.less(row_index, col_index),
+            torch.greater_equal(col_index, num_input_tokens),
+        ),
+        torch.greater_equal(row_index, num_input_tokens),
+    )
+    mask = torch.logical_not(mask)
+    mask = mask.unsqueeze(0).unsqueeze(0)
+    return mask
+
+  def forward(self, input_tokens, time_step):
+    # input_tokens: [1, T]
+    # time_step: []
+    paddings = torch.sum(input_tokens == 0, dim=-1)
+    paddings = paddings[0]
+    _, T = input_tokens.shape  # pylint: disable=invalid-name
+
+    num_paddings = torch.sum(paddings)
+    num_input_tokens = T - num_paddings
+
+    orig_mask = self.build_mask(num_input_tokens + time_step)
+    orig_mask = orig_mask[:, :, time_step : time_step + T, :]
+
+    new_mask = orig_mask[:, :, :, time_step : time_step + T].clone()
+    zeroed_mask = torch.zeros_like(new_mask)
+
+    orig_mask[:, :, :, time_step : time_step + T] = zeroed_mask
+
+    mask = torch.cat([orig_mask, new_mask], dim=-1)
+
+    return {'mask': mask}