slundberg

4 ms·

1. 1.▲
   by slundberg 3y ago

   You can serialize and ship the DSL to a remote server for high speed execution. (without trusting raw Python code)
2. 2.▲
   by slundberg 3y ago

   As others mentioned, this was initially developed before LangChain became widely used. Since it is lower level, you can leverage other tools, like any vector store interface you like such as in LangChain. Writing complex chain of thought st
3. 3.▲
   by slundberg 3y ago

   If you want guidance acceleration speedups (and token healing) then you have to use an open model locally right now, though we are working on setting up a remote server solution as well. I expect APIs will adopt some support for more contro
4. 4.▲
   Show HN: Interpretable Machine Learning with XGBoost(towardsdatascience.com)

   1 points by slundberg 8y ago|0 comments
5. 5.▲
   Show HN: A unified approach to explain the output of any machine learning model(github.com)

   1 points by slundberg 9y ago|0 comments