Its my second week exploring RASA. But I found that all the tutorials that supposedly make rasa popular such as, botfront, rasa-demo, rasa-x, rasa-webchat no longer work with the default and latest release/version of rasa. From various blogs and forums, and repos, I have seen a lot of struggles by developers trying to make rasa to rapidly develop solutions. From what I learned exploring it in the internet and with firsthand experience myself, it was useful covering the years 2020-2021, but 2022 to 2023, all the supposedly major frameworks that supported it ceased support and developers are looking for quick answers especially when experiencing broken dependencies from non-working repos.
My question is, for instance we have the following which I have repeatedly visited coming from search results (even results from you.com [powered by an older version of rasa] are not entirely useful. The docs are too hard to follow so that, I need to consult ChatGPT and YOU.com to clarify the meaning or compatibility of certain examples about RASA supposedly famous 2 or 3 years ago, but can never be worked out today. On that note, is there any topic other than the following below that describes about an updated version of this tutorial?
I think my last question is one of concern: What strategies do rasa community have to expand simplicity on the usefulness of their opensource product? I was trying to find the answers for the last 2 weeks.
Disappointing examples and the feeling of discouragement led me to huggingface’s GPT and it turns out that even YOU.COM (powered by rasa) and ChatGPT (powered by GPT3) could answer best about questions on GPT rather than rasa. Now, (with the help of these AIs) I was able to launch GPT2 on a cloud server ready to serve business clients about its potential. I don’t want to implement service from the apis provided by openai’s GPT3, but a custom one. I was looking for a way to make it possible with RASA, but all web searches returns frustrations and my deepest questions remain un-answered, unlike pytorch’s support for machine learning which is the easiest to implement and deploy online.
I hope this can change. I feel like I’m in the dark and I believe I am representing a lot of voices in the developer community. My first goal was to develop rasa interaction with GPT2 (at least on my first 3 days in machine learning about two weeks ago) and I still need some light on this subject since I’d open to giving rasa open source a fair try.
I think rasa needs to show more concrete working examples with its latest release that works, the same way that they used to work 2 or 3 years ago. I still go to the rasa docs from time to time, but humans need examples to learn quickly and developers wanted a quick solution to certain problems.
Please direct me with some guidance that can bring light to what I like to achieve.
Thank you very much.