Fine Tuning T5 for comment generation task

Hi everyone!
I wanted to fine-tune text generation models such as T5 on a task of conditional comment generation. An example is below :

Linkedin post author: Contentful
Linkedin post:
Today’s digital content can & should be reusable. To do this, modular content should be used to solve scaling issues through smart content reuse. Read the recent Forrester report to learn more.
Linkedin post comments:
[Laurel Plimpton] comment: Is this a good resource?
Reply to [Laurel Plimpton] comment:

Output: Yes, it is a great resource. It explains how to use modular content to solve scaling issues and also offers tips on how to create reusable digital content.

I was confused about whether providing everything as the context and asking the model to generate the comments would work or not! GPT-3 is able to generate good responses but I want to achieve the same with open source models. Can anyone help me with how can I fine-tune T5 or any relevant model for this task?

Thank you!