Transformer-smaller-training-vocab

What's Changed
* fix handling of added special tokens in tokenizers by helpmefindaname in https://github.com/helpmefindaname/transformer-smaller-training-vocab/pull/6


**Full Changelog**: https://github.com/helpmefindaname/transformer-smaller-training-vocab/compare/0.2.2...0.2.3

What's Changed
* add support for torch 2.0.0 by helpmefindaname in https://github.com/helpmefindaname/transformer-smaller-training-vocab/pull/5


**Full Changelog**: https://github.com/helpmefindaname/transformer-smaller-training-vocab/compare/0.2.1...0.2.2

What's Changed
* Fix saving of reduced models: When the models are saved while being reduced, they now properly set the vocab-size in the config, allowing the used to load the model again, no matter if the context manager is still there or not. helpmefindaname in https://github.com/helpmefindaname/transformer-smaller-training-vocab/pull/4
* If the embeddings are frozen (not trainable), the reduced embeddings will also be frozen helpmefindaname in https://github.com/helpmefindaname/transformer-smaller-training-vocab/pull/4


**Full Changelog**: https://github.com/helpmefindaname/transformer-smaller-training-vocab/compare/0.2.0...0.2.1

What's Changed
* Update Optimizer Parameters by helpmefindaname in https://github.com/helpmefindaname/transformer-smaller-training-vocab/pull/3

Introduces an optional parameter `optimizer` to `reduce_train_vocab`. Which can be used to modify the parameter group to exchange changed embeddings to the new ones. An example usage is the following:


model = ...
tokenizer = ...
optimizer = Adam(model.parameters(), lr=...)
...

with reduce_train_vocab(model=model, tokenizer=tokenizer, texts=get_texts_from_dataset(raw_datasets, key="text"), optimizer=optimizer):
train_with_optimizer(model, tokenizer, optimizer)

save_model() save model at the end to contain the full vocab again.


**Full Changelog**: https://github.com/helpmefindaname/transformer-smaller-training-vocab/compare/0.1.8...0.2.0

What's Changed
* Lower dependency requirements to transformers 4.1, torch 1.8 and datasets 2.0 as the package was previously too restrictive


**Full Changelog**: https://github.com/helpmefindaname/transformer-smaller-training-vocab/compare/0.1.7...0.1.8

**Full Changelog**: https://github.com/helpmefindaname/transformer-smaller-training-vocab/compare/0.1.6...0.1.7