SynapseML v0.11, a distributed open-source machine learning library, has been released.
- It includes new features such as OpenAI language models and embeddings, simple deep learning, LightGBM v2 , ONNX Model Hub, and causal learning.
- OpenAI language models enable the creation of complex LLM prompts from data frames, OpenAI embeddings allow quick and easy embedding of tables of sentences, and the OpenAIChatCompletion transformer enables thousands of conversations to be inferred in parallel.
- Simple deep learning makes it possible to train custom text and deep vision classifiers with only a few lines of code.
- LightGBM v2 provides high performance through new streaming APIs.
- ONNX Model Hub is a collection of state-of-the-art pre-trained ONNX models.
- Finally, causal learning helps make informed decisions by establishing causation, and SynapseML's causal learning package implements the "Double machine learning" technique.
