- Sort Score
- Result 10 results
- Languages All
Results 21 - 30 of 306 for Quantized (0.11 sec)
-
tensorflow/compiler/mlir/quantization/stablehlo/quantization_config.proto
// previous quantized layer (Please note that this part is different part // from DEBUGGER_TYPE_FLOAT_PER_LAYER). Each layer in the debugging model // has a DumpTensor, and it is used to save the entire value of outputs from // both the quantized and unquantized layer. DEBUGGER_TYPE_INT_PER_LAYER = 2; // DEBUGGER_TYPE_FLOAT_PER_LAYER creates a debugging model with both
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Fri May 17 03:36:50 UTC 2024 - 14.3K bytes - Viewed (0) -
tensorflow/compiler/mlir/quantization/stablehlo/cc/report.cc
return result; } else { return std::nullopt; } } // Populates quantized ops from `module_op` to `results`. After going through // the quantization passes, quantized ops are represented as `func::CallOp` with // a callee's prefix of `quantized_`. void PopulateQuantizedResults(ModuleOp module_op, QuantizationResults& results) {
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 6.8K bytes - Viewed (0) -
tensorflow/compiler/mlir/quantization/tensorflow/tests/quantize_composite_functions.mlir
// CHECK: -------- Quantization Summary -------- // CHECK: Number of quantized layers in the model // CHECK: -------------------------------- // CHECK: Name Count/Total // CHECK: ================================ // CHECK: Conv2D 1/2 // CHECK: Number of quantized layers with quantized outputs: 1/1 // CHECK: Number of quantize layers added: 1 // CHECK: Number of dequantize layers added: 1 } // -----
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Mon Nov 06 01:23:21 UTC 2023 - 15.2K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/stablehlo/transforms/uniform_quantized_stablehlo_to_tfl_pass.cc
// * Input tensors are per-tensor uniform quantized (i8->f32) // tensors (full integer) with shape [..., r_x, c_x] or [..., c_x, r_x]. // * The filter tensor is a per-tensor uniform quantized (i8->f32) tensor // (constant or activation) with shape [..., r_y, c_y] or [..., c_y, r_y]. // * Output tensors are per-tensor uniform quantized (i8->f32) or // per-channel uniform quantized (i32->f32) tensors. //
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Mon Apr 22 09:00:19 UTC 2024 - 99.8K bytes - Viewed (0) -
tensorflow/compiler/mlir/quantization/stablehlo/passes/quantization_patterns.h
namespace mlir::quant::stablehlo { // Checks whether an op is connected with a quantized composite function. If // not, the same-scale op will not be quantized. This decision is based on the // current assumption that the performance gain of the same-scale op itself // could not beat the overhead of the quantize and dequantize routines need to // be added around that op. When the assumption changes, this policy might
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 10.9K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/stablehlo/transforms/passes.td
def ComposeUniformQuantizedTypePass : Pass<"compose-uniform-quantized-type", "ModuleOp"> { let summary = "Compose uniform quantized types in StableHLO."; let constructor = "mlir::odml::CreateComposeUniformQuantizedTypePass()"; let description = [{ Identifies uniform quantization patterns and composes them to uniform quantized types. This pass targets a specific set of models that are quantized from the framework level, which produces "decomposed"
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 21:59:06 UTC 2024 - 5.6K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/transforms/prepare_quantize.cc
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 17.6K bytes - Viewed (0) -
tensorflow/compiler/mlir/quantization/stablehlo/tests/components/post_calibration_component.mlir
// RUN: stablehlo-quant-opt %s -stablehlo-test-post-calibration-component='unpack-quantized-types=false' \ // RUN: -split-input-file | FileCheck %s --check-prefix=CHECK-NO-UNPACK // Tests that a simple dot_general (lifted as a function) with CustomAggregators // around it is quantized. The resulting graph has quantized types unpacked into // int ops. func.func @main(%arg0: tensor<1x1024xf32>) -> tensor<1x3xf32> {
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 01:09:50 UTC 2024 - 6.7K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/quantization/ir/QuantOps.td
// quantized representation may be acceptable. // // Especially early in transformation, it is common to have pairs of // qcast/dcast at points where a transition to a quantized type is // required. In addition, it is also common to have an identity qcast // (where the operand and result type are not quantized) at all points where // it is legal to use a quantized representation (but is not known to be
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Oct 13 12:46:08 UTC 2022 - 10.2K bytes - Viewed (0) -
tensorflow/compiler/mlir/quantization/stablehlo/cc/report_test.cc
// The quantized call op without the _quantization_method attribute is not // captured as a `QuantizationResult`. ASSERT_THAT(results.results(), IsEmpty()); } TEST_F(QuantizationReportTest, InitializeWithModuleOpWithInvalidCalleeName) { // A quantized dot_general op but the callee function has an invalid name. It // is expected to start with `quantized_`.
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 10:10:34 UTC 2024 - 18.5K bytes - Viewed (0)