v0.0.24

What's changed

@wanoz

• updated readme intro sections for added clarity
• updated setup.py

v0.0.23

What's changed

@wanoz

• fixed bug where input argument for predict function was not working with list variables
• updated setup.py

v0.0.22

What's changed

@wanoz

• added models of falcon-7b and 40b variants
• updated readme with HuggingFace Open LLM leaderboard
• updated setup.py

v0.0.21

What's changed

@wanoz

• added more details in readme
• updated setup.py

v0.0.20

What's changed

@wanoz

• fixed huggingface predict not taking in the transform argument
• updated setup.py
• updated readme

v0.0.19

What's changed

@wanoz

• refactor: move lookup values into constants.py
• added support for RedPajama models
• updated setup.py
• fixed type hinting of predict functions
• change max_tokens to max_length in direct argument input for OpenAI for better argument name consistency in the library
• added the ability for prompt modifier to take in a function
• fixed setup.py readme.md file read encoding

v0.0.18

What's changed

@wanoz

• refined GPU status message
• added support for mac m-series chip GPU
• added support for batch prediction on dataframe column
• updated readme and docstring
• fixed input typing check bug
• fixed seq2seq training with evaluation and corrected trainingargument

v0.0.17

What's changed

@wanoz

• Fixed issue with handling model_args GPU setting variable
• Added related unit tests

v0.0.16

What's changed

@wanoz

• adding prompt templates
• added future annotation import for supprting Python 3.8 or lower
• added GPU setting for training and inference for HuggingFace models
• updated setup.py

v0.0.15

What's changed

@Hanshan1988:

• Fixed error in flan t5 embeddings
• Added HuggingFace Hub model support for bert, distilbert, roberta and distilroberta base models