Add ORT inference (#113)
* added gpu extras and added > transformers for token-classification pipeline issue
* added numpy and huggingface hub to required packages
* added modeling_* classes
* adding tests and pipelines
* remove vs code folder
* added test model and adjusted gitignore
* add readme for tests
* working tests
* added some documentation
* will ci run?
* added real model checkpoints
* test ci
* fix styling
* fix some documentation
* more doc fixes
* added some feedback and wording from michael and lewis
* renamed model class to ORTModelForXX
* moved from_transformers to from_pretrained
* applied ellas feedback
* make style
* first version of ORTModelForCausalLM without past-keys
* added first draft of new .optimize method
* added better quantize method
* fix import
* remove optimize and quantize
* added lewis feedback
* added style for test
* added >>> to code snippets
* style
* added condition for staging tests
* feedback morgan & michael
* added action
* forgot to install pytest
* forgot sentence piece
* made sure we won't have import conflicts
* make style happy
