* attemping to deprecate python 3.5 and onboard 3.9. Can't wait to see what breaksgit add .
* Holding on 3.9 for multiple deps
* making 3.6 also run TVM
* native multioutput regression support for some sklearn models
* fixing numerical precision issues
* working around numerical precision issues
* wip
* minor changes
* adding xgb missing value support for tree_trav methods
* xgb missing value support for gemm tree implementation
* fixing error failures
* fixing more errors
* fixing tvm issue
* missing value can be encoded as anything
* adding missing value handling support for LightGBM
* adding both classification and regression tests
* add max string param to operators working on strings
* Add native support for string features
* fix tests to use the new string type
* adding supported string types to constants
* [WIP] onnxml Label Encoder
* adding another case for nfeatures
* fix to label encoder
* add onnx feature vectorizer
Co-authored-by: Karla Saur <kasaur@microsoft.com>
* [WIP] onnxml Label Encoder
* lint
* adding another case for nfeatures
* fix to label encoder
* fixing nfeatures from rebase
* adding failcase
* checking in progress
* adding test for bad backend intput
* adding skl labelenc
* rebase to #387
* tmp fix so my other tests will also run...
* cleanup
* getting the right error for bad input
* adding failing case for onnxml strings
* fix to label encoder
* fix error
* try with ravel
* adding failcase for onnx strings
* typo
* cleanup
* add test for onnxml label encoder on string
fix bug in convert when inputs have len(shape) == 1
* add comment on the limit of onnx export for label encoder string
* bumping to torch==1.8 for some cases
* RuntimeError for olderpyt
* removing extra line
* saving 1.8 for another day
Co-authored-by: Matteo Interlandi <mainterl@microsoft.com>
* introducing BatchContainer
* BatchContainer basic functionality done
* pass test_input to _convert
* introduce convert_batch API
* use convert_batch in the benchmark
* store _batch_size attribute
* test working
* run black, add concat output option, fix benchmark
* fix getattr
* fix operator benchmark
* support transform and decision function
* make sure input is tuple not list
* fix torch backend prediction
* begin fixing tests
* squeeze and ravel on onnx regression output
* all tests in test_extra_conf.py working
* restore BATCH_SIZE and k neighbor test
* fix onnxml test
* run black on test_extra_conf.py
* fix test_sklearn_normalizer_converter.py
* fix test_lightgbm_converter.py
* fixing more onnxml tests
* fixed remaining onnxml tests
* use format, fix pylint
* fix typo
* add document
* add missing doc
* fix typo
* doc update, remove unused stuff
* trying to get pipeline working
* updating string
* rtol atol
* pinning to macos version, reverint to 1 cache. removing rtol for now..
* macos 10.14
* rtol back
* Fix problem with inputs of shape == 0
Use proper data types for scalers
Remove ONNX_INPUTS since is not used
* Add missing test
fix problem with N_features when no test input is passed
* add explisti case for concat
Karla Saur
Matteo Interlandi
Matteo Interlandi
Supun Nakandala
qin-xiong
masahi