* Changes to fuse embed layer for gpt2, kernal changes pending
* verified add output and regular add match
* Test added for additional output embedlayernorm, working on CUDA
* Test passing on CPU
* updated convert_to_onnx toll to check parity correctly
* removed some debugs
* couple of TODO left as in optimizer.py
* removed changes to optimizer.py
* fixing build
* fixing build
* updated order of initilization
* added a test case for float16
* updating the docs
* updating tests failing due to embed layer fusion
* update unit tests
* updating CUDA documentation in operatorkernels.md
* addressing comments
* OperatorKernels.md updated with CUDA
* adding TODO to qembed_layer
* minor edit
* updated docs
* addressing comments
* adding position ids to embed layer gpt2
* updating fused gpt2 model
* added extra test
* remove comments
* addressing comments
* contrib_defs.cc updated
* all tests passing
* fixing a typo
* minor edit
* trigger build
* qembedlayernorm checkinputs updated
* fixing build error
* fixing build error
* fixing build error
* add initializer checker for Gather with 1D input
* Check if indices value exists
* Update symbolic_shape_infer.py
* add unit test
* Update symbolic_shape_infer.py
* Update symbolic_shape_infer.py
* implement cuda provider
* define profiler common
* call start after register
* add memcpy event
* add cuda correlation
* format code
* add cupti to test path
* switch to CUpti_ActivityKernel3
* reset cupti path
* fix test case
* fix trt pipeline
* add namespace
* format code
* exclude training from testing
* remove mutex
(1) Attention Fusion for gpt-2 model from Megatron.
(2) Update symbolic shape inference of Attention to support 4D mask.
(3) Add an otpion in save_model_to_file to save external data in one file or not, and warning of existing external data
(4) Fix deprecation: logger.warn => logger.warning
(5) Add model loader to test model without external data
(6) Add an API of optimize_by_fusion, and topological sort after optimization.
* Include pytorch_export_contrib_ops in inference builds
Rename / move it from tools/python/register_custom_ops_pytorch_exporter
to onnxruntime/python/tools/pytorch_export_contrib_ops.
Rationale for inclusion in inference builds:
This code is potentially useful for anyone using ORT, not just training.
Rationale for new name:
"Contrib op" is the nomenclature used within ORT to refer to the set of
ops that are not in the standard op set but are included by default with
ORT. This is more specific than "custom op", which is what the PyTorch
exporter uses to refer to any non-standard op.
Step 1 of addressing #8818. After this is merged I will update the docs.
* Enable test_pytorch_export_contrib_ops.py in CI
Fixes AB#1342330
* support register external ep lib inforation; make eager mode share the same ep pools with training workloads
* fix inference code
* fix build break
* fix the message
Allow FastGelu half2 kernel to build without --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=xx
Add environment variable ORT_TRANSFORMER_OPTIONS=4 to disable half2 FastGelu kernel for testing purpose
Test parity of FastGelu operator with fp16 inputs.
* rename BertOptimizationOptions to FusionOptions
* remove disable_onnxruntime, and use opt_level to control whether onnxruntime graph optimization is used.
* Change default opt_level for backward compatible. When opt_level is not specified, default value is based on model type.
Re-enable tests that disabled in PR 8530
Update import of test_optimizer.py so that the test could run in source directory.
Add a parameter to disable symbolic shape inference in fp16 conversion since it throws exception for some model.
* enable shared lib test on linux
* fix build break
* add onnx dependency
* add rpath
* skip the test for linux training
* set ONNX_ML definition
* install training python dependency
* update
* fix format; add eigen include folder
* fix format
* skip amd build
* enable shared provider on training
* fix comments in pr
Co-authored-by: Ubuntu <chenta@chenta-orttraining-cpu.bxgbzpva45kedp3rhbsbit4phb.jx.internal.cloudapp.net>
Co-authored-by: Changming Sun <chasun@microsoft.com>
SparseTensor support
Implement Builder pattern
Fix support for 1-D and 2-D COO indices
Implement and test CSR support.
Handle shape inference for SparseTensors
Implement conversion for COO, CSR and tests.
Address the case where constant sparse initializer is the output.
Implement test infra for SparseTensors
Implement SparseDenseMatMul for Csr and COO and tested it.
Add hash for SparseToDenseMatMul
Finish shared provider refactor
Refactor GetOrCreate to Create
Working on py interface
Expose OrtDevice and use it in allocate_numpy
Adjust Sparse interfaces, add support for string SparseTensor. Add tests.
Add and test to_cuda()
Add accessors to format specific indices
Test values and indices views, read-only flag, after GC access
Add sparse related methods to OrtValue
Re-work SparseTensor wrapper, add OrtValue methods
Rework numpy_array_to_cuda/to_cpu
Add run_with_ort_values
Add models and test sparse_mat_mul with run_with_ort_values
Refactor sparse tensor to use a single buffer
Ifdef x86 Eigen CSR sparse matmul implementation
Exclude broken test, check for string type when copying cross device
Split pybind schema, regenerate docs, add exclusion
Conditionally exclude schema module
Update docs fix cuda build
Add test to a filter and renerate JS docs
Add conversion and test string support for sparse tensors
Exclude conversion utils from minimal build
Add CUDA Memcpy and adjust provider interfaces
* changes working to convert akv nodes
* changes to replace nodes
* changes to accomodate qkv hidden sizes as attributes
* kernel to accept qkv_hidden_size attributes
* Working till compute for varied dimension, todo applyattention()
* changes to make all regression tests work
* inference running successfully without prepack
* success inference with pre-pack weights
* add test for diff sizes
* bias shape need not be a mul of 3
* get the output_hidden_size from input
* infer output shape from input
* merge with master
* cleaning up files that got merged wrong
* accurancy at accepted level
* added unit test case for different dimensions
* all unit tests passing
* packed weights working for attention
* prepacked weights working
* added test case for newly added extra qk input
* updated unit test to test only extra add qk
* fixing build error
* removing few debugs
* reverting test changes
* all python test passing
* cleaning up
* new unit test added, major clean up of code
* removed extra code
* minor
* minor fix to tests
* prepack weights code cleaned up
* compacted compute() in attention.cc
* reformat compute()
* making a parameter T
* adding 3 q,k,v buffers in all cases
* fixing build
* running tests only on cpu
* Updating docs
* trigger ci builds
* Addressing comments in PR
* addressing some more comments
* get add_qk_str from add_qk node directly
* updating docs, added extra check to verify attn inputs
* Optimized the extra add by parallelizing
* added attention_shape to symbolic_shape_infer.py
* minor refactoring to address comments
* changes to fuse attention node and create varied dimensions
* added an option to optimizer to only do offline fusion
* fixing a typo
* merge with master
* removing extra changes
* added new unit test - test_attention_fusion_for_varied_qkv_dimensions()
* Unit test succesfull for q,k,v paths with varied dimensions
* adding test model for unit test case
* optimizing attention tests
* removing debugs
* minor change
* addressing comments
* addressing comments
* changed the new option to disable_onnxruntime
* replacing asserts with debugs
* make attn fusion backward compatible for head_size, hidden_size
* preserving behavior for shape_modified_tensor
* adding new option as the last parameter
* cleaning up
* line breaks and spaces
* formatting according to python
* making the changes to fuse attention node without user input
* changes to fusion_attention.py updated
* bringing the code up to python standard
* checkin transformers pipeline
* add docker requirements
* only trigger linux cpu
* temp remove tf instalation due to numpy version conflicts
* test numpy>=1.7
* revert numpy and disable transformers
* add coloredlogs
* enable shape_infer_helper and install transformers when needed
* pip3?
* testtest
* enable more tets
* line too long
* remove pytorch1.4 test and added back some onnx files
* add tests
* copy dir
* disable 2 teests
* trim lines
* add missing onnx
* fix type
* fix version conflicts
* install psutil
* change file path
* mfix path
* remove cached files
* add back attention fusion test
* labeled the shape infer test as slow
* fix
* enable tf2onnx test and enable pytest
* refactor path
* fix typo
* add cwd
* First iteration of making cuda a shared provider.
Separated out shared OpKernel change, so doing this to merge with that change.
* More cuda shared library refactoring
* More cuda shared library refactoring
* More build options tested, converted the training ops over.
* Fix merge breaks
* Fix submodules
* Fix submodules
* Fix submodules
* Fix python
* Fix compile errors
* Duplicate symbol fix
* Test fix for ROCM provider
* Another ROCM test workaround
* ROCM Build Test
* ROCM build fix
* ROCM
* ROCM
* ROCM
* ROCM
* ROCM
* ROCM test
* Reduce header dependencies
* Remove redundant namespace
* Test fix for linux
* Fix linux build
* Fix Eigen build error
* Fix unused parameter warning
* Test link error
* Another linker test
* Linker test
* Linker test
* Another test
* Another build test
* Fix linux link error
* Build test
* Fix control flow ops to use common base class with core code
* Remove extra qualifiers
* Fix template syntax for linux
* Fix cuda memory leak
* Fix pybind
* Test disabling cast
* Cleanup
* Restore cuda in test
* Remove more header dependencies
* Test not adding cuda provider to session
* Make GetProviderInfo_CUDA throw
* No-op cuda provider creation
* Fix some setup issues
* Fix memory cleanup on unload
* Diagnostics
* Don't unload library
* Add diagnostics
* Fix deleting registry at right time.
* Test disabling profiler
* Fix merge break
* Revert profiler change
* Move unloading of shared providers into Environment
* Free more global allocations before library unloads
* Add more diagnostics
* Move unloading back to the OrtEnv as there are multiple Environments created during a session.
Remove some library dependencies for tests.
* Fix more cmake files
* ERROR -> WARNING
* Fix python shutdown
* Test not using dml in pipeline
* Change python version and disable dml
* Update python version
* Test adding unload method for shared providers
* Disable DLL test
* Python test
* Revert "Python test"
This reverts commit c7ec2cfe98.
* Revert "Disable DLL test"
This reverts commit e901cb93aa.
* Revert "Test adding unload method for shared providers"
This reverts commit c427b78799.
* Point to RyanWinGPU
* Revert python version
* Fix id_to_allocator_map
* Another python exit test
* Remove extra debug messages
Try a more clean python shutdown through DllMain
* Revert DllMain idea, it didn't work
* Merge conflicts
* Fix merge with master issues.
* Comments
* Undo edit to file
* Cleanup + new training ops
* Revert yml changes
* Fix another merge error
* ROCM fix
* ROCM fix v2
* Put back Linux hack, it is necessary
* Stupid fixes
* Fix submodule out of sync
* ROCM fix 3
* ROCM 4
* Test java fix
* Fix typos
* Java test on my VM
* Fix build error
* Spotless fix
* Leave temp file around to load properly
* Fix cleanup on exit
* Fix break
* Java comments
* Remove LongformerAttentionBase workaround
* Spotless fix
* Switch yml back to regular build pool
* Revert "Switch yml back to regular build pool"
This reverts commit be35fc2a5a.
* Code review feedback
* Fix errors due to merge
* Spotless fix
* Fix minimal build
* Java fix for non cuda case
* Java fix for CPU build
* Fix Nuphar?
* Fix nuphar 2
* Fix formatting
* Revert "Remove LongformerAttentionBase workaround"
This reverts commit 648679b370.
* Training fix
* Another java fix
* Formatting
* Formatting
* For orttraining
* Last orttraining build fix...
* training fixes
* Fix test provider error
* Missing pass command
* Removed in wrong spot
* Python typo
* Python typos
* Python crash on exit, possibly due to unloading of libraries.
* Remove test_execution_provider from training build
Only enable python atexit on windows
Remove assert on provider library exit
* Still can't unload providers in python, alas.
* Disable Nvtx temporarily
* MPI Kernels for Training
* MPI Kernels part 2
* Patch through INcclService
* Oops, wrong CMakeLists
* Missing namespace
* Fix missing ()
* Move INcclService::GetInstance around to link nicer
* Missing }
* Missing MPI libraries for Cuda
* Add extra GetType functions used by MPI
* Missing Nccl library
* Remove LOGS statements as a test
* Add in a couple more missing GetType methods
* Update comments
* Missed a logging reference in mpi_context.h
* Convert aten_op to shared (due to marge with master)
* Test moving DistributedRunContext instance into shared provider layer
(with purpose error to verify it's being built properly)
* Test passed, now with fix
* Missing static
* Oops, scope DistributedRunContext to just NCCL
* Merge related issues and code review feedback.
* Merge error
* Bump to rel-1.9.1 (#7684)
* Formatting
* Code review feedback for Java build on non Windows
* Remove cupti library dependency from core library
* Test Java pipeline fix
* Linux build fix
* Revert "Linux build fix"
This reverts commit a73a811516.
* Revert "Remove cupti library dependency from core library"
This reverts commit 6a889ee8bf.
* Packaging pipeline fixes to copy cuda shared provider for tensorrt & standard packages
* Add cuda to Tensorrt nuget package
* onnxruntime_common still has a cuda header dependency
Co-authored-by: ashbhandare <ash.bhandare@gmail.com>
* Add FBGEMM submodule
* Add fbgemm based per-channel quantization
* Add missing logic for pre-layernorm transformer model fusion
* add support for structured pruning architecture -fastformers
* Fix windows build
* Add a default behavior when head_size is not present for the backward compatibility
* Remove FBGEMM and default to tensor-wise quantization, column-wise quantization will be enabled later
* Fixed some unit test errors
* Fix windows compile error and unit test errors
* delete the option removed from the upstream
* Addresses review comments and fixes a merge error
* Remove commented out code
* add non-zero zp support
* support A and B scale with any dimensions
* fix build breaks
* fix warning in MSVC
* Fix bug for not checking original float value names when treat it as not existing.
* Clean up head size
* Clean up python tools
* Enable per column quantization
* fix quant weight cleanup bug
* A few code clean up
* Some code clean-up
* Some code clean-up
* Change option name
* update default value
* Rename option and parameter names
* Missing argument name change
* Add tests for quantization options for attention and matmul
Co-authored-by: Yufeng Li <liyufeng1987@gmail.com>
Co-authored-by: Lei Zhang <zhang.huanning@hotmail.com>
