Improve code generation & Fine tuning the model
zernonia opened this issue · comments
I realised that the current multipass approach from https://openv0.com/ might not be the best.
Problem:
- After generating the code, we didnt let ChatGPT know if the generated code is /
- The result for follow up iteration might be very different from previous version.
- The generation is taking quite (as compare to v0)
More reading and research need to be done!! Would love to hear your thoughts! 😁