Model Fine-tuning and Optimized Prompt Templates
Hi there,
Thank you for sharing the model.
Currently, I am utilizing instruction prompts from the GitHub repository located at https://github.com/google-research/FLAN/blob/main/flan/v2/flan_templates_branched.py for my conversational chatbot.
I am curious if the model has been fine-tuned using specific sets of optimized prompt templates or structures. If there are any accompanying write-ups or blog posts detailing the process of fine-tuning the model from flan-t5-xl, I would greatly appreciate it if you could share them with me.
Lastly, I would be grateful if you could provide the optimized prompt templates/structures to enhance the text generation capabilities of the model, ensuring optimal performance and quality.
I look forward to hearing from you.
Best regards.
This is the default prompt for fastchat-t5
https://github.com/lm-sys/FastChat/blob/f60b66093ca08c44701af00114df92d3ab66f3ac/fastchat/conversation.py#L251-L282