Comment by nzoschke

I have tried it many times and I want to like it but I find the LLM doesn't program in it well.

Maybe it's my fault, or maybe a model training, or something to improve in docs and SDKs?