Ollama models are not respecting the system prompt / context

[lfnovo]

When using Ollama as the chat models, for some reason, they are not recognizing the context in the system prompt, rendering them useless for the Chat purpose.

I plan to investigate more on this issue soon, but if you are familiar with using Ollama with Langchain, I'd appreciate your help. For the time being, I recommend using other providers for chat.

Ollama models work great for transformations and can be used with no issues there.