onnxruntime-directml 1.24.1__cp314-cp314-win_amd64.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- onnxruntime/LICENSE +21 -0
- onnxruntime/Privacy.md +21 -0
- onnxruntime/ThirdPartyNotices.txt +6121 -0
- onnxruntime/__init__.py +418 -0
- onnxruntime/backend/__init__.py +6 -0
- onnxruntime/backend/backend.py +175 -0
- onnxruntime/backend/backend_rep.py +52 -0
- onnxruntime/capi/DirectML.dll +0 -0
- onnxruntime/capi/__init__.py +4 -0
- onnxruntime/capi/_ld_preload.py +7 -0
- onnxruntime/capi/_pybind_state.py +33 -0
- onnxruntime/capi/build_and_package_info.py +2 -0
- onnxruntime/capi/convert_npz_to_onnx_adapter.py +48 -0
- onnxruntime/capi/onnxruntime.dll +0 -0
- onnxruntime/capi/onnxruntime_collect_build_info.py +47 -0
- onnxruntime/capi/onnxruntime_inference_collection.py +1440 -0
- onnxruntime/capi/onnxruntime_providers_shared.dll +0 -0
- onnxruntime/capi/onnxruntime_pybind11_state.pyd +0 -0
- onnxruntime/capi/onnxruntime_validation.py +154 -0
- onnxruntime/capi/version_info.py +2 -0
- onnxruntime/datasets/__init__.py +18 -0
- onnxruntime/datasets/logreg_iris.onnx +0 -0
- onnxruntime/datasets/mul_1.onnx +0 -0
- onnxruntime/datasets/sigmoid.onnx +13 -0
- onnxruntime/quantization/CalTableFlatBuffers/KeyValue.py +78 -0
- onnxruntime/quantization/CalTableFlatBuffers/TrtTable.py +90 -0
- onnxruntime/quantization/CalTableFlatBuffers/__init__.py +0 -0
- onnxruntime/quantization/__init__.py +19 -0
- onnxruntime/quantization/base_quantizer.py +529 -0
- onnxruntime/quantization/calibrate.py +1267 -0
- onnxruntime/quantization/execution_providers/qnn/__init__.py +2 -0
- onnxruntime/quantization/execution_providers/qnn/fusion_lpnorm.py +132 -0
- onnxruntime/quantization/execution_providers/qnn/fusion_spacetodepth.py +162 -0
- onnxruntime/quantization/execution_providers/qnn/mixed_precision_overrides_utils.py +413 -0
- onnxruntime/quantization/execution_providers/qnn/preprocess.py +353 -0
- onnxruntime/quantization/execution_providers/qnn/quant_config.py +389 -0
- onnxruntime/quantization/fusions/__init__.py +4 -0
- onnxruntime/quantization/fusions/fusion.py +311 -0
- onnxruntime/quantization/fusions/fusion_gelu.py +272 -0
- onnxruntime/quantization/fusions/fusion_layernorm.py +146 -0
- onnxruntime/quantization/fusions/replace_upsample_with_resize.py +96 -0
- onnxruntime/quantization/matmul_bnb4_quantizer.py +239 -0
- onnxruntime/quantization/matmul_nbits_quantizer.py +1638 -0
- onnxruntime/quantization/neural_compressor/__init__.py +1 -0
- onnxruntime/quantization/neural_compressor/onnx_model.py +1251 -0
- onnxruntime/quantization/neural_compressor/util.py +80 -0
- onnxruntime/quantization/neural_compressor/weight_only.py +932 -0
- onnxruntime/quantization/onnx_model.py +600 -0
- onnxruntime/quantization/onnx_quantizer.py +1163 -0
- onnxruntime/quantization/operators/__init__.py +2 -0
- onnxruntime/quantization/operators/activation.py +119 -0
- onnxruntime/quantization/operators/argmax.py +18 -0
- onnxruntime/quantization/operators/attention.py +73 -0
- onnxruntime/quantization/operators/base_operator.py +26 -0
- onnxruntime/quantization/operators/binary_op.py +72 -0
- onnxruntime/quantization/operators/concat.py +62 -0
- onnxruntime/quantization/operators/conv.py +260 -0
- onnxruntime/quantization/operators/direct_q8.py +78 -0
- onnxruntime/quantization/operators/embed_layernorm.py +121 -0
- onnxruntime/quantization/operators/gather.py +64 -0
- onnxruntime/quantization/operators/gavgpool.py +62 -0
- onnxruntime/quantization/operators/gemm.py +172 -0
- onnxruntime/quantization/operators/lstm.py +121 -0
- onnxruntime/quantization/operators/matmul.py +231 -0
- onnxruntime/quantization/operators/maxpool.py +34 -0
- onnxruntime/quantization/operators/norm.py +40 -0
- onnxruntime/quantization/operators/pad.py +172 -0
- onnxruntime/quantization/operators/pooling.py +67 -0
- onnxruntime/quantization/operators/qdq_base_operator.py +22 -0
- onnxruntime/quantization/operators/resize.py +34 -0
- onnxruntime/quantization/operators/softmax.py +74 -0
- onnxruntime/quantization/operators/split.py +63 -0
- onnxruntime/quantization/operators/where.py +87 -0
- onnxruntime/quantization/preprocess.py +141 -0
- onnxruntime/quantization/qdq_loss_debug.py +389 -0
- onnxruntime/quantization/qdq_quantizer.py +1477 -0
- onnxruntime/quantization/quant_utils.py +1051 -0
- onnxruntime/quantization/quantize.py +953 -0
- onnxruntime/quantization/registry.py +110 -0
- onnxruntime/quantization/shape_inference.py +204 -0
- onnxruntime/quantization/static_quantize_runner.py +256 -0
- onnxruntime/quantization/tensor_quant_overrides.py +520 -0
- onnxruntime/tools/__init__.py +10 -0
- onnxruntime/tools/check_onnx_model_mobile_usability.py +47 -0
- onnxruntime/tools/convert_onnx_models_to_ort.py +380 -0
- onnxruntime/tools/file_utils.py +47 -0
- onnxruntime/tools/logger.py +11 -0
- onnxruntime/tools/make_dynamic_shape_fixed.py +73 -0
- onnxruntime/tools/mobile_helpers/__init__.py +0 -0
- onnxruntime/tools/mobile_helpers/coreml_supported_mlprogram_ops.md +53 -0
- onnxruntime/tools/mobile_helpers/coreml_supported_neuralnetwork_ops.md +43 -0
- onnxruntime/tools/mobile_helpers/nnapi_supported_ops.md +58 -0
- onnxruntime/tools/mobile_helpers/usability_checker.py +738 -0
- onnxruntime/tools/offline_tuning.py +169 -0
- onnxruntime/tools/onnx_model_utils.py +416 -0
- onnxruntime/tools/onnx_randomizer.py +85 -0
- onnxruntime/tools/onnxruntime_test.py +164 -0
- onnxruntime/tools/optimize_onnx_model.py +56 -0
- onnxruntime/tools/ort_format_model/__init__.py +27 -0
- onnxruntime/tools/ort_format_model/operator_type_usage_processors.py +653 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/__init__.py +0 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/ArgType.py +7 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/ArgTypeAndIndex.py +67 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/Attribute.py +337 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/AttributeType.py +18 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/Checkpoint.py +125 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/DeprecatedKernelCreateInfos.py +120 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/DeprecatedNodeIndexAndKernelDefHash.py +68 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/DeprecatedSessionState.py +96 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/DeprecatedSubGraphSessionState.py +72 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/Dimension.py +71 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/DimensionValue.py +80 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/DimensionValueType.py +8 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/EdgeEnd.py +32 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/FloatProperty.py +67 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/Graph.py +320 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/InferenceSession.py +88 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/IntProperty.py +67 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/KernelTypeStrArgsEntry.py +91 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/KernelTypeStrResolver.py +78 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/MapType.py +71 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/Model.py +223 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/ModuleState.py +141 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/Node.py +317 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/NodeEdge.py +126 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/NodeType.py +7 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/NodesToOptimizeIndices.py +160 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/OpIdKernelTypeStrArgsEntry.py +91 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/OperatorSetId.py +67 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/OptimizerGroup.py +117 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/ParameterOptimizerState.py +91 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/PropertyBag.py +152 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/RuntimeOptimizationRecord.py +105 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/RuntimeOptimizationRecordContainerEntry.py +91 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/RuntimeOptimizations.py +79 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/SequenceType.py +58 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/Shape.py +78 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/SparseTensor.py +114 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/StringProperty.py +67 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/StringStringEntry.py +67 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/Tensor.py +203 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/TensorDataType.py +26 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/TensorTypeAndShape.py +71 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/TypeInfo.py +83 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/TypeInfoValue.py +9 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/ValueInfo.py +84 -0
- onnxruntime/tools/ort_format_model/ort_flatbuffers_py/fbs/__init__.py +6 -0
- onnxruntime/tools/ort_format_model/ort_model_processor.py +86 -0
- onnxruntime/tools/ort_format_model/types.py +85 -0
- onnxruntime/tools/ort_format_model/utils.py +61 -0
- onnxruntime/tools/pytorch_export_contrib_ops.py +129 -0
- onnxruntime/tools/pytorch_export_helpers.py +131 -0
- onnxruntime/tools/qdq_helpers/__init__.py +0 -0
- onnxruntime/tools/qdq_helpers/optimize_qdq_model.py +37 -0
- onnxruntime/tools/qnn/add_trans_cast.py +292 -0
- onnxruntime/tools/qnn/gen_qnn_ctx_onnx_model.py +364 -0
- onnxruntime/tools/qnn/preprocess.py +165 -0
- onnxruntime/tools/reduced_build_config_parser.py +203 -0
- onnxruntime/tools/remove_initializer_from_input.py +37 -0
- onnxruntime/tools/symbolic_shape_infer.py +3094 -0
- onnxruntime/tools/update_onnx_opset.py +31 -0
- onnxruntime/transformers/__init__.py +8 -0
- onnxruntime/transformers/affinity_helper.py +40 -0
- onnxruntime/transformers/benchmark.py +942 -0
- onnxruntime/transformers/benchmark_helper.py +643 -0
- onnxruntime/transformers/bert_perf_test.py +629 -0
- onnxruntime/transformers/bert_test_data.py +641 -0
- onnxruntime/transformers/compare_bert_results.py +256 -0
- onnxruntime/transformers/constants.py +47 -0
- onnxruntime/transformers/convert_generation.py +3605 -0
- onnxruntime/transformers/convert_tf_models_to_pytorch.py +205 -0
- onnxruntime/transformers/convert_to_packing_mode.py +385 -0
- onnxruntime/transformers/dynamo_onnx_helper.py +205 -0
- onnxruntime/transformers/float16.py +501 -0
- onnxruntime/transformers/fusion_attention.py +1189 -0
- onnxruntime/transformers/fusion_attention_clip.py +340 -0
- onnxruntime/transformers/fusion_attention_sam2.py +533 -0
- onnxruntime/transformers/fusion_attention_unet.py +1307 -0
- onnxruntime/transformers/fusion_attention_vae.py +300 -0
- onnxruntime/transformers/fusion_bart_attention.py +435 -0
- onnxruntime/transformers/fusion_base.py +141 -0
- onnxruntime/transformers/fusion_bias_add.py +57 -0
- onnxruntime/transformers/fusion_biasgelu.py +66 -0
- onnxruntime/transformers/fusion_biassplitgelu.py +110 -0
- onnxruntime/transformers/fusion_conformer_attention.py +222 -0
- onnxruntime/transformers/fusion_constant_fold.py +144 -0
- onnxruntime/transformers/fusion_embedlayer.py +810 -0
- onnxruntime/transformers/fusion_fastgelu.py +492 -0
- onnxruntime/transformers/fusion_gelu.py +258 -0
- onnxruntime/transformers/fusion_gelu_approximation.py +25 -0
- onnxruntime/transformers/fusion_gemmfastgelu.py +121 -0
- onnxruntime/transformers/fusion_gpt_attention.py +546 -0
- onnxruntime/transformers/fusion_gpt_attention_megatron.py +355 -0
- onnxruntime/transformers/fusion_gpt_attention_no_past.py +260 -0
- onnxruntime/transformers/fusion_group_norm.py +180 -0
- onnxruntime/transformers/fusion_layernorm.py +489 -0
- onnxruntime/transformers/fusion_mha_mmdit.py +667 -0
- onnxruntime/transformers/fusion_nhwc_conv.py +99 -0
- onnxruntime/transformers/fusion_options.py +340 -0
- onnxruntime/transformers/fusion_qordered_attention.py +420 -0
- onnxruntime/transformers/fusion_qordered_gelu.py +118 -0
- onnxruntime/transformers/fusion_qordered_layernorm.py +122 -0
- onnxruntime/transformers/fusion_qordered_matmul.py +216 -0
- onnxruntime/transformers/fusion_quickgelu.py +74 -0
- onnxruntime/transformers/fusion_reshape.py +173 -0
- onnxruntime/transformers/fusion_rotary_attention.py +1591 -0
- onnxruntime/transformers/fusion_shape.py +109 -0
- onnxruntime/transformers/fusion_simplified_layernorm.py +165 -0
- onnxruntime/transformers/fusion_skip_group_norm.py +254 -0
- onnxruntime/transformers/fusion_skiplayernorm.py +209 -0
- onnxruntime/transformers/fusion_transpose.py +167 -0
- onnxruntime/transformers/fusion_utils.py +321 -0
- onnxruntime/transformers/huggingface_models.py +74 -0
- onnxruntime/transformers/import_utils.py +20 -0
- onnxruntime/transformers/io_binding_helper.py +487 -0
- onnxruntime/transformers/large_model_exporter.py +395 -0
- onnxruntime/transformers/machine_info.py +230 -0
- onnxruntime/transformers/metrics.py +163 -0
- onnxruntime/transformers/models/bart/__init__.py +12 -0
- onnxruntime/transformers/models/bart/export.py +98 -0
- onnxruntime/transformers/models/bert/__init__.py +12 -0
- onnxruntime/transformers/models/bert/eval_squad.py +329 -0
- onnxruntime/transformers/models/gpt2/__init__.py +12 -0
- onnxruntime/transformers/models/gpt2/benchmark_gpt2.py +413 -0
- onnxruntime/transformers/models/gpt2/convert_to_onnx.py +566 -0
- onnxruntime/transformers/models/gpt2/gpt2_helper.py +1031 -0
- onnxruntime/transformers/models/gpt2/gpt2_parity.py +513 -0
- onnxruntime/transformers/models/gpt2/gpt2_tester.py +501 -0
- onnxruntime/transformers/models/gpt2/parity_check_helper.py +146 -0
- onnxruntime/transformers/models/llama/__init__.py +12 -0
- onnxruntime/transformers/models/llama/benchmark.py +700 -0
- onnxruntime/transformers/models/llama/benchmark_all.py +488 -0
- onnxruntime/transformers/models/llama/benchmark_e2e.py +608 -0
- onnxruntime/transformers/models/llama/convert_to_onnx.py +1064 -0
- onnxruntime/transformers/models/llama/dist_settings.py +57 -0
- onnxruntime/transformers/models/llama/llama_inputs.py +504 -0
- onnxruntime/transformers/models/llama/llama_parity.py +343 -0
- onnxruntime/transformers/models/llama/llama_torch.py +47 -0
- onnxruntime/transformers/models/llama/quant_kv_dataloader.py +108 -0
- onnxruntime/transformers/models/longformer/__init__.py +12 -0
- onnxruntime/transformers/models/longformer/benchmark_longformer.py +821 -0
- onnxruntime/transformers/models/longformer/convert_to_onnx.py +413 -0
- onnxruntime/transformers/models/longformer/generate_test_data.py +347 -0
- onnxruntime/transformers/models/longformer/longformer_helper.py +76 -0
- onnxruntime/transformers/models/phi2/__init__.py +12 -0
- onnxruntime/transformers/models/phi2/convert_to_onnx.py +590 -0
- onnxruntime/transformers/models/phi2/inference_example.py +414 -0
- onnxruntime/transformers/models/sam2/__init__.py +12 -0
- onnxruntime/transformers/models/sam2/benchmark_sam2.py +638 -0
- onnxruntime/transformers/models/sam2/convert_to_onnx.py +270 -0
- onnxruntime/transformers/models/sam2/image_decoder.py +272 -0
- onnxruntime/transformers/models/sam2/image_encoder.py +236 -0
- onnxruntime/transformers/models/sam2/mask_decoder.py +208 -0
- onnxruntime/transformers/models/sam2/nvtx_helper.py +33 -0
- onnxruntime/transformers/models/sam2/prompt_encoder.py +189 -0
- onnxruntime/transformers/models/sam2/sam2_demo.py +321 -0
- onnxruntime/transformers/models/sam2/sam2_image_onnx_predictor.py +279 -0
- onnxruntime/transformers/models/sam2/sam2_utils.py +147 -0
- onnxruntime/transformers/models/stable_diffusion/__init__.py +12 -0
- onnxruntime/transformers/models/stable_diffusion/benchmark.py +1519 -0
- onnxruntime/transformers/models/stable_diffusion/benchmark_controlnet.py +426 -0
- onnxruntime/transformers/models/stable_diffusion/demo_txt2img.py +103 -0
- onnxruntime/transformers/models/stable_diffusion/demo_txt2img_xl.py +269 -0
- onnxruntime/transformers/models/stable_diffusion/demo_utils.py +778 -0
- onnxruntime/transformers/models/stable_diffusion/diffusion_models.py +1318 -0
- onnxruntime/transformers/models/stable_diffusion/diffusion_schedulers.py +1179 -0
- onnxruntime/transformers/models/stable_diffusion/engine_builder.py +295 -0
- onnxruntime/transformers/models/stable_diffusion/engine_builder_ort_cuda.py +387 -0
- onnxruntime/transformers/models/stable_diffusion/engine_builder_ort_trt.py +288 -0
- onnxruntime/transformers/models/stable_diffusion/engine_builder_tensorrt.py +395 -0
- onnxruntime/transformers/models/stable_diffusion/engine_builder_torch.py +108 -0
- onnxruntime/transformers/models/stable_diffusion/optimize_pipeline.py +590 -0
- onnxruntime/transformers/models/stable_diffusion/ort_optimizer.py +136 -0
- onnxruntime/transformers/models/stable_diffusion/pipeline_stable_diffusion.py +831 -0
- onnxruntime/transformers/models/stable_diffusion/trt_utilities.py +12 -0
- onnxruntime/transformers/models/t5/__init__.py +12 -0
- onnxruntime/transformers/models/t5/convert_to_onnx.py +318 -0
- onnxruntime/transformers/models/t5/t5_decoder.py +437 -0
- onnxruntime/transformers/models/t5/t5_encoder.py +70 -0
- onnxruntime/transformers/models/t5/t5_encoder_decoder_init.py +361 -0
- onnxruntime/transformers/models/t5/t5_helper.py +302 -0
- onnxruntime/transformers/models/whisper/__init__.py +12 -0
- onnxruntime/transformers/models/whisper/benchmark.py +585 -0
- onnxruntime/transformers/models/whisper/benchmark_all.py +526 -0
- onnxruntime/transformers/models/whisper/convert_to_onnx.py +609 -0
- onnxruntime/transformers/models/whisper/whisper_chain.py +334 -0
- onnxruntime/transformers/models/whisper/whisper_decoder.py +464 -0
- onnxruntime/transformers/models/whisper/whisper_encoder.py +164 -0
- onnxruntime/transformers/models/whisper/whisper_encoder_decoder_init.py +371 -0
- onnxruntime/transformers/models/whisper/whisper_helper.py +1035 -0
- onnxruntime/transformers/models/whisper/whisper_inputs.py +380 -0
- onnxruntime/transformers/models/whisper/whisper_jump_times.py +477 -0
- onnxruntime/transformers/onnx_exporter.py +719 -0
- onnxruntime/transformers/onnx_model.py +1636 -0
- onnxruntime/transformers/onnx_model_bart.py +141 -0
- onnxruntime/transformers/onnx_model_bert.py +488 -0
- onnxruntime/transformers/onnx_model_bert_keras.py +474 -0
- onnxruntime/transformers/onnx_model_bert_tf.py +588 -0
- onnxruntime/transformers/onnx_model_clip.py +42 -0
- onnxruntime/transformers/onnx_model_conformer.py +32 -0
- onnxruntime/transformers/onnx_model_gpt2.py +101 -0
- onnxruntime/transformers/onnx_model_mmdit.py +112 -0
- onnxruntime/transformers/onnx_model_phi.py +929 -0
- onnxruntime/transformers/onnx_model_sam2.py +137 -0
- onnxruntime/transformers/onnx_model_t5.py +985 -0
- onnxruntime/transformers/onnx_model_tnlr.py +226 -0
- onnxruntime/transformers/onnx_model_unet.py +258 -0
- onnxruntime/transformers/onnx_model_vae.py +42 -0
- onnxruntime/transformers/onnx_utils.py +55 -0
- onnxruntime/transformers/optimizer.py +620 -0
- onnxruntime/transformers/past_helper.py +149 -0
- onnxruntime/transformers/profile_result_processor.py +358 -0
- onnxruntime/transformers/profiler.py +434 -0
- onnxruntime/transformers/quantize_helper.py +76 -0
- onnxruntime/transformers/shape_infer_helper.py +121 -0
- onnxruntime/transformers/shape_optimizer.py +400 -0
- onnxruntime/transformers/torch_onnx_export_helper.py +74 -0
- onnxruntime_directml-1.24.1.dist-info/METADATA +216 -0
- onnxruntime_directml-1.24.1.dist-info/RECORD +322 -0
- onnxruntime_directml-1.24.1.dist-info/WHEEL +5 -0
- onnxruntime_directml-1.24.1.dist-info/entry_points.txt +2 -0
- onnxruntime_directml-1.24.1.dist-info/top_level.txt +1 -0
|
@@ -0,0 +1,31 @@
|
|
|
1
|
+
#!/usr/bin/env python3
|
|
2
|
+
# Copyright (c) Microsoft Corporation. All rights reserved.
|
|
3
|
+
# Licensed under the MIT License.
|
|
4
|
+
|
|
5
|
+
import argparse
|
|
6
|
+
import os
|
|
7
|
+
import pathlib
|
|
8
|
+
|
|
9
|
+
from .onnx_model_utils import update_onnx_opset
|
|
10
|
+
|
|
11
|
+
|
|
12
|
+
def update_onnx_opset_helper():
|
|
13
|
+
parser = argparse.ArgumentParser(
|
|
14
|
+
f"{os.path.basename(__file__)}:{update_onnx_opset_helper.__name__}",
|
|
15
|
+
description="""
|
|
16
|
+
Update the ONNX opset of the model.
|
|
17
|
+
New opset must be later than the existing one.
|
|
18
|
+
If not specified will update to opset 15.
|
|
19
|
+
""",
|
|
20
|
+
)
|
|
21
|
+
|
|
22
|
+
parser.add_argument("--opset", type=int, required=False, default=15, help="ONNX opset to update to.")
|
|
23
|
+
parser.add_argument("input_model", type=pathlib.Path, help="Provide path to ONNX model to update.")
|
|
24
|
+
parser.add_argument("output_model", type=pathlib.Path, help="Provide path to write updated ONNX model to.")
|
|
25
|
+
|
|
26
|
+
args = parser.parse_args()
|
|
27
|
+
update_onnx_opset(args.input_model, args.opset, args.output_model)
|
|
28
|
+
|
|
29
|
+
|
|
30
|
+
if __name__ == "__main__":
|
|
31
|
+
update_onnx_opset_helper()
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
# -------------------------------------------------------------------------
|
|
2
|
+
# Copyright (c) Microsoft Corporation. All rights reserved.
|
|
3
|
+
# Licensed under the MIT License.
|
|
4
|
+
# --------------------------------------------------------------------------
|
|
5
|
+
import os
|
|
6
|
+
import sys
|
|
7
|
+
|
|
8
|
+
sys.path.append(os.path.dirname(__file__))
|
|
@@ -0,0 +1,40 @@
|
|
|
1
|
+
# -------------------------------------------------------------------------
|
|
2
|
+
# Copyright (c) Microsoft Corporation. All rights reserved.
|
|
3
|
+
# Licensed under the MIT License.
|
|
4
|
+
# --------------------------------------------------------------------------
|
|
5
|
+
|
|
6
|
+
# Get/Set cpu affinity. Currently only support part of Unix system
|
|
7
|
+
import logging
|
|
8
|
+
import os
|
|
9
|
+
|
|
10
|
+
logger = logging.getLogger(__name__)
|
|
11
|
+
|
|
12
|
+
|
|
13
|
+
class AffinitySetting:
|
|
14
|
+
def __init__(self):
|
|
15
|
+
self.pid = os.getpid()
|
|
16
|
+
self.affinity = None
|
|
17
|
+
self.is_os_supported = hasattr(os, "sched_getaffinity") and hasattr(os, "sched_setaffinity")
|
|
18
|
+
if not self.is_os_supported:
|
|
19
|
+
logger.warning("Current OS does not support os.get_affinity() and os.set_affinity()")
|
|
20
|
+
|
|
21
|
+
def get_affinity(self):
|
|
22
|
+
if self.is_os_supported:
|
|
23
|
+
self.affinity = os.sched_getaffinity(self.pid)
|
|
24
|
+
|
|
25
|
+
def set_affinity(self):
|
|
26
|
+
if self.is_os_supported:
|
|
27
|
+
current_affinity = os.sched_getaffinity(self.pid)
|
|
28
|
+
if self.affinity != current_affinity:
|
|
29
|
+
logger.warning(
|
|
30
|
+
"Replacing affinity setting %s with %s",
|
|
31
|
+
str(current_affinity),
|
|
32
|
+
str(self.affinity),
|
|
33
|
+
)
|
|
34
|
+
os.sched_setaffinity(self.pid, self.affinity)
|
|
35
|
+
|
|
36
|
+
|
|
37
|
+
if __name__ == "__main__":
|
|
38
|
+
affi_helper = AffinitySetting()
|
|
39
|
+
affi_helper.get_affinity()
|
|
40
|
+
affi_helper.set_affinity()
|