- Sort Score
- Result 10 results
- Languages All
Results 11 - 20 of 130 for created (0.45 sec)
-
tensorflow/compiler/mlir/tf2xla/internal/clustering_bridge_passes.cc
// Guarantee all functions have one use, which enables more exact shape // inference. pm.addPass(mlir::TF::CreateGuaranteeAllFuncsOneUsePass()); // Run shape inference so that tf_executor/tf_device ops created later will // likely to inherit more concrete types. pm.addPass(mlir::TF::CreateTFShapeInferencePass()); pm.addNestedPass<FuncOp>(mlir::TFTPU::CreateTPUPartitionedOpConversionPass()); pm.addNestedPass<FuncOp>(
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Tue Apr 30 16:09:14 UTC 2024 - 11.2K bytes - Viewed (0) -
tensorflow/compiler/mlir/tf2xla/transforms/legalize_tf_communication.cc
return tokens[0]; } else { return builder.create<AfterAllOp>(loc, original_token.getType(), tokens) .getResult(); } } // Replaces `tf._XlaHostComputeMlir` with individual `mhlo.send` and `mhlo.recv` // ops per operand and result. Unique Channel IDs are assigned per transfer. // Sink tokens are created across all `mhlo.send` ops first and then by // all `mhlo.recv` ops.
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 40.5K bytes - Viewed (0) -
tensorflow/compiler/jit/xla_launch_util.cc
if (pjrt_tensor_buffer != nullptr) { args->push_back(pjrt_tensor_buffer->pjrt_buffer()); } else { // Creates a PjRtBuffer from DeviceMemoryBase. The newly created // PjRtBuffer needs to be persisted till XLA execution is completed. auto dmem = se::DeviceMemoryBase( const_cast<char*>(tensor->tensor_data().data()),
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu May 16 00:36:08 UTC 2024 - 40.4K bytes - Viewed (0) -
tensorflow/compiler/mlir/tf2xla/internal/passes/extract_head_tail_outside_compilation.cc
} // Wraps block in a Launch. External uses of ops in the block will be return // values of the Launch and remapped to the Launch results. If `before` is set // to true, the Launch is created before `op`. Otherwise the Launch is created // after `op`. mlir::tf_device::LaunchOp CreateLaunchForBlock(OpBuilder* builder, Operation* op, bool before,
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 20.6K bytes - Viewed (0) -
tensorflow/compiler/jit/increase_dynamism_for_auto_jit_pass_test.cc
Out(NodeWith( Op("Slice"), Name("begin/static_shaped_slice/static_shaped_slice"))), _))); } // New constants being created need to have control dependencies copied to // ensure correct control flow analysis in TF V2. TEST(SliceToDynamicSliceRewriteTest, WithControlDepsToConstant) { Scope root = Scope::NewRootScope()
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Feb 22 08:47:20 UTC 2024 - 18.3K bytes - Viewed (0) -
tensorflow/c/eager/gradients.cc
void MarkAsResult(AbstractTensorHandle* gradient) const override; void DeleteGradient(AbstractTensorHandle* gradient) const override; private: // The context where the aggregation op `Add` is to be created. AbstractContext* ctx_; }; // Returns the number of elements in the gradient tensor. int64_t TapeVSpace::NumElements(AbstractTensorHandle* tensor) const {
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Feb 15 09:49:45 UTC 2024 - 19.3K bytes - Viewed (0) -
tensorflow/compiler/mlir/tensorflow/transforms/functional_control_flow_to_regions.cc
} void runOnOperation() override; }; // Creates a call to function `func` in region `caller_region`. Use `args` as // the call arguments, and terminate the region with a yield. The arguments are // cast to the required type before the call. `use_region_args` control whether // the input arguments are used as is (for IfOp) or block arguments of the same // type as the input arguments are created and then used as call arguments (for // While).
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 11K bytes - Viewed (0) -
tensorflow/compiler/mlir/tensorflow/transforms/cluster_tf_ops_pass.cc
builder.clone(*op, mapping); } // Creates the ReturnOp so that the per-host function returns the // correct values of the cloned operations. llvm::SmallVector<Value, 4> results_after_mapping; for (Value result : metadata.results) { results_after_mapping.push_back(mapping.lookupOrDefault(result)); } builder.create<func::ReturnOp>(loc, results_after_mapping);
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 13.4K bytes - Viewed (0) -
tensorflow/compiler/jit/xla_device.cc
// Caches a XlaDeviceAllocator per <backend, device ordinal> pair. A // XlaDeviceAllocator is created on demand and is associated with a // XlaDevice. It outlives the device itself (for instance, the buffer // backing a tensor holds a pointer to the allocator for book-keeping, // and this buffer can outlast the device). class XlaDeviceAllocatorState { public: // Creates or returns a cached XlaDeviceAllocator for a given // backend and device_ordinal.
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Mon May 20 21:05:42 UTC 2024 - 24.3K bytes - Viewed (0) -
tensorflow/compiler/mlir/quantization/stablehlo/passes/bridge/convert_tf_quant_types.cc
!IsQintValueDefinedByIntToQintCast(op->getOperand(i))) { new_operands.push_back(rewriter.create<TF::CastOp>( op->getLoc(), orig_op_type, operands[i])); } else { new_operands.push_back(operands[i]); } } // Create a new UQ op. OperationState state(op->getLoc(), op->getName().getStringRef(),
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 12.6K bytes - Viewed (0)