update .gitignore

support for transformers 4.41.0

Merge pull request #60 from McGill-NLP/split-datasets-diff-instructions

Process split instructions in code rather than data files

Merge pull request #37 from McGill-NLP/supervised-training

Supervised training

MNTP training (#15)

* first iteration of script ,TODO: save properly

* model class loading

* bf16 flash attn

* ignore wandb

* added dependencies for running mlm training

* revert custom local change

* peft loading and saving

* remove unused functions

* all configs

* minor formatting changes

* flash attention needs no-build-isolation

Merge pull request #12 from McGill-NLP/bi-models

Bi models

Update README.md

Update README.md

Updated readme