Palashio Libra Versions Save

• added fine_tune parameter that allows layer freezing in CNN's
• fixes for custom architecture
• plots and classification report for CNN
• dashboard fixes
• nlp fixes

Just some minor bug fixes all around!

New features:

• Working full fledged dashboard
• Deploy models to TensorFlow Lite
• Use DCGAN's with a gan query
• Plots are now available for CNN query
• Added more pretrained models for CNN queries: DenseNet, MobileNet etc.
• Quite a few bug changes

• Fixed issues with pprint dependency
• Adjusted path for dashboard

• small fixes for named entity recognition

• addition of XGBoost Query
• addition of reccomendation engine
• addition of dashboard(), which is a fully functional UI for data analysis and dimensionality reduction integration.
• addition of text generation query
• ability to use pretrained models in convolutional() query
• named entity recognition query addition

• Added XGBoost Support
• Added interactive dataset UI
• Fixed issues with python-Levenshetein import
• Removal of torch dependencies - complete reimplementation in tensorflow!

Contributors

@Abhilash2000 @ramyabuva @jbofill10 @dikshantj @abx393 @vraj123

Fixed issues with increasing layers after every loss calculation. Increased optimizations using list comprehensions.

Contributors

@Palashio @Sar1hak @Abhilash2000

Fixed issue with wrong accuracy display for classification_ANN query. Also removed redundant code for neural network creation for neural_network_query().

Contributors

@Sar1hak @Abhilash2000 @Palashio

First official release of libra! This build includes robust implementations of our core features.

Contributors

Contributions from these developers are featured in this build:

@ramyabuva @SidAkalwadi @pahuja-gor @vyathakavilocana @yash19062000 @jbofill10 @Sar1hak @anas-awadalla @pranavnt @Pragun100 @Vakhshoori101