Currently the llm generates text continuing from // JSON:
This means that chain of thought is not properly implemented as the llm is not given the opportunity to generate 'thought'.
Fix json errors and formatting
There are syntax / formatting errors which from quick testing tend to confused the llm in terms of the required format.
Allow LLM to process reasoning (chain of thought)
Currently the llm generates text continuing from
// JSON:This means that chain of thought is not properly implemented as the llm is not given the opportunity to generate 'thought'.Fix json errors and formatting
There are syntax / formatting errors which from quick testing tend to confused the llm in terms of the required format.
Note
I have other suggestions as outlined in: https://discourse.mcneel.com/t/ghpt-generates-a-grasshopper-script-from-a-prompt-with-chat-gpt/162039/22?u=samgregson
However here I have implemented only those suggestions which seem like simple mistakes and leave suggestions for separate PRs