- Sort Score
- Result 10 results
- Languages All
Results 11 - 16 of 16 for FakeQuant (0.12 sec)
-
tensorflow/compiler/mlir/quantization/common/quantization_lib/quantization_utils.cc
llvm::DenseSet<Operation*> redundant_stats_ops; // Step 0: remove the quantfork::StatisticsOp which are used by the // quant.qcast op in case it overrides the information from training FakeQuant // ops. func.walk([&](quantfork::QuantizeCastOp q) { auto input_op = q.getArg().getDefiningOp(); if (auto stats = dyn_cast_or_null<quantfork::StatisticsOp>(input_op)) { q.setOperand(stats.getArg());
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Wed May 08 02:10:16 UTC 2024 - 43.2K bytes - Viewed (0) -
tensorflow/compiler/mlir/quantization/common/quantization_lib/quantization_utils.h
// outputs. llvm::DenseMap<SignedInteger, QuantParamsForResults> restricted_output_params; // Coefficient operand index and whether supporting per-channel quantization. // For QAT, this information is carried by the FakeQuant*/Quantize/Dequantize // ops, but post-training quantization, the quantization parameters need to be // inferred from the tensor content and op property. A "-1" value indicates
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Wed Apr 24 20:30:06 UTC 2024 - 41.7K bytes - Viewed (0) -
tensorflow/compiler/mlir/tensorflow/transforms/lower_tf.cc
values[output_index] = item; } } rewriter.replaceOpWithNewOp<PackOp>(op, op.getType(), values); return success(); } }; // This pass performs a manual conversion with FakeQuant, converting between // floating point and quantized space. It is designed to reproduce TF's // implementation, mirroring the previous XLA implementation. //
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Apr 25 16:01:03 UTC 2024 - 74.9K bytes - Viewed (0) -
tensorflow/compiler/mlir/quantization/tensorflow/python/integration_test/quantize_model_test.py
loader = saved_model_loader.SavedModelLoader(self._output_saved_model_path) graphdef = loader.get_meta_graph_def_from_tags(tags).graph_def # The Conv2D op shouldn't be quantized as it has no FakeQuant on input. self.assertTrue(self._contains_op(graphdef, 'Conv2D')) # If the Gather op is quantized, input_model_size / output_model_size > 2. self.assertLess( testing.get_size_ratio(
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Fri May 17 03:36:50 UTC 2024 - 235.6K bytes - Viewed (0) -
tensorflow/compiler/mlir/lite/ir/tfl_ops.td
let results = (outs TFL_FpTensor:$output); } def TFL_FakeQuantOp : TFL_Op<"fake_quant", [ Pure, QuantizableResult]> { let summary = "FakeQuant operator"; let description = [{ Fake-quantize the 'inputs' tensor of type float via float scalars min and max to 'outputs' tensor of same shape as inputs. }]; let arguments = (
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Thu Jun 06 19:09:08 UTC 2024 - 186K bytes - Viewed (0) -
RELEASE.md
tensorflow is running in. * It is now safe to call any of the C API's TF_Delete\* functions on nullptr. * Log some errors on Android to logcat. * Match FakeQuant numerics in TFLite to improve accuracy of TFLite quantized inference models. * Optional bucket location check for the GCS Filesystem. * Performance enhancements for StringSplitOp & StringSplitV2Op.
Registered: Sun Jun 16 05:45:23 UTC 2024 - Last Modified: Tue Jun 11 23:24:08 UTC 2024 - 730.3K bytes - Viewed (0)