forrestbao

5 ms·

1. 1.▲
   by forrestbao 2y ago

   This rocks! An LLM tailored for RAG.
2. 2.▲
   by forrestbao 3y ago

   Thank you very much! Appreciate your comment and encouragement. I conceived this idea while working on a project with multiple people to evaluate multiple LLMs. Too many API keys floating here and there in the code. So I said, what if we co
3. 3.▲
   Show HN: LlaMaKey – One master key for all cloud LLM/GenAI APIs(github.com)

   6 points by forrestbao 3y ago|2 comments
4. 4.▲
   by forrestbao 3y ago

   A smaller model has less capacity and thus is less prone to overfitting, which is a reason of hallucination. Overfitting means that a model cannot adjust its output based on the input that is unseen during training.
5. 5.▲
   Fast and neat Arduino prototyping in BreadShield. No more jumpwiring(crowdsupply.com)

   2 points by forrestbao 7y ago|0 comments
6. 6.▲
   BreadShield: Bridging Arduino to breadboard without jumper wires(crowdsupply.com)

   2 points by forrestbao 7y ago|0 comments