- Sort Score
- Result 10 results
- Languages All
Results 1 - 2 of 2 for TPUCopyWithDynamicShapeOp (0.23 sec)
-
tensorflow/compiler/mlir/g3doc/_includes/tf_passes.md
}) : () -> tensor<f32> return %0 : tensor<f32> } ``` ### `-tf-extract-tpu-copy-with-dynamic-shape-op` _Extract the TPUCopyWithDynamicShapeOp out of the host launch and place it on device launch_ This pass looks for TPUCopyWithDynamicShapeOp which wraps in a `tf_device.launch` with host device attribute. It extracts the ops and wrap them in `tf_device.launch` with tpu device attribute so that ops can be
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Wed Aug 02 02:26:39 UTC 2023 - 96.4K bytes - Viewed (0) -
tensorflow/compiler/mlir/tensorflow/ir/tf_ops.td
} def TF_TPUAnnotateTensorsWithDynamicShapeOp : TF_Op<"TPUAnnotateTensorsWithDynamicShape", [Pure]> { let summary = [{ Placeholder op which takes the output of TPUCopyWithDynamicShapeOp and pass them to the following tpu ops. }]; let description = [{ This op serves as an annotation for the dynamic shaped tensor and will be removed during the bridge rewrite. }];
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Wed Apr 24 04:08:35 UTC 2024 - 90.5K bytes - Viewed (0)