Textwiser

Latest version: v2.0.1

Safety actively analyzes 683322 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 2 of 3

1.3.1

-------------------------------------------------------------------------------
minor:
- Updated dependency requirements to reflect updates to Flair library

-------------------------------------------------------------------------------

1.3.0

-------------------------------------------------------------------------------
major:
- Added T5, XLM-RoBERTa, BART, ECELTRA, DialoGPT, Longformer embeddings
- Added inline pooling for Word embeddings to make some pooling scenarios faster

minor:
- Updated basic usage notebook
- Updated preferred method of persistence & examples
- SVD now properly appears in model parameters and can be fine-tuned

-------------------------------------------------------------------------------

1.2.1

-------------------------------------------------------------------------------

minor:
- Fix transformers model initialization for newer versions of transformers

-------------------------------------------------------------------------------

1.2.0

-------------------------------------------------------------------------------
major:
- Update Universal Sentence Encoder
- Now works with newer versions
- Default version is now v5
- Add bytepair embeddings
- Add ALBERT embeddings

minor:
- Add explicit fit_transform method to Compound embeddings to make them faster

-------------------------------------------------------------------------------

1.1.1

-------------------------------------------------------------------------------

minor:
- Fix running transformers models on a GPU
- Fix modification of input schema object for Compound embeddings

-------------------------------------------------------------------------------

1.1.0

-------------------------------------------------------------------------------

major:
- All transformer-based models are powered by the transformers library
- This enables fine-tuning for transformer models!
- There isn't a subword to word level aggregation, everything is aggregated at doc level
- The models output the last layer by default
- The scalar mix parameter is discontinued
- Add DistilBERT and CTRL models
- Add Sphinx documentation
- Add LICENSE

minor:
- More modular word embedding testing

-------------------------------------------------------------------------------

Page 2 of 3

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.