RasaGPT: First headless LLM chatbot built on top of Rasa, Langchain and FastAPI

I&#x27;m not sure what the advantage the use of a somewhat comprehensive framework like Langchain gives you for this use case?<p>It starts to feel as AI tech is slowly turning into web tech with a million tools and frameworks, so I&#x27;m just wondering whether all of these are needed and if it isn&#x27;t easier to code your own than learning a foreign framework...

Unfortunately there were not a whole lot of end-to-end examples of integrating Rasa with OpenAI nor functional boilerplates on github so I put a working prototype together in a few days and thus RasaGPT was bron.<p>RasaGPT is a python-based boilerplate and reference implementation of Rasa and Telegram utilizing an LLM library like Langchain for indexing, retrieval and context injection. FastAPI end-points are made available for you to build your application on top of. Features include:<p>- Automated hand-off to human if queries are out of bounds - &quot;Training&quot; pipeline done via API - Multi-tenant support - Generate category labels from questions - Works right out of the box with docker-compose - Ngrok reverse tunnel and dummy data included - Multiple use cases and a great starting point<p>Hope you like it, more @ rasagpt.dev

A bit off topic but you better change the name and remove the GPT. OpenAI is claiming AI products that are using GPT in their name are causing confusion and is sending legal threats now.<p>One of many examples: <a href="https:&#x2F;&#x2F;twitter.com&#x2F;pbteja1998&#x2F;status&#x2F;1654095756200931328" rel="nofollow">https:&#x2F;&#x2F;twitter.com&#x2F;pbteja1998&#x2F;status&#x2F;1654095756200931328</a>

Sorry for the off topic question, but does anyone know how to buy consumer hardware optimal for running emerging open source chat models with the largest parameter chat models possible?<p>Would it be more cost effective to try to buy an absurd amount of ram and run on the cpu?<p>Or buy an Nvidia card with the biggest capacity available?<p>Or maybe buy a Mac with the most memory you can get?

Also, with Haystack and a smaller Transformer model to address the long-tail of answers <a href="https:&#x2F;&#x2F;github.com&#x2F;deepset-ai&#x2F;rasa-haystack">https:&#x2F;&#x2F;github.com&#x2F;deepset-ai&#x2F;rasa-haystack</a> (and <a href="https:&#x2F;&#x2F;www.deepset.ai&#x2F;blog&#x2F;build-smart-conversational-agents-with-chatbots-qa" rel="nofollow">https:&#x2F;&#x2F;www.deepset.ai&#x2F;blog&#x2F;build-smart-conversational-agent...</a>)

Tangentially, it&#x27;s interesting seeing an open source project like this actually spin up a domain name, contact email, and some branding (the image in the Readme), for a project the author said was created in just a few days.<p>I wonder what the objective is for that extra polish. If it&#x27;s optimizing star count growth, how much do these touches help?

Curious if people want to suggest alternatives to Rasa for writing stateful chatbots. Or share feedback about using Rasa.

Can somebody ELI5 Rasa for me? I read through the README and I still don&#x27;t get what it does.

Can you actually build a reliable customer-facing chatbot on top of LLM&#x27;s? With the &quot;jailbreaking&quot; and not knowing if it&#x27;s actually using the data you&#x27;re supplying it or other data it was trained on and so on.

Everybody racing into the AI space to plant their flag and say &quot;First!&quot;. But first isn&#x27;t going to be correlated with the winner much, I&#x27;d wager

what problem is this solving?

I&#x27;d suggest to not put a bunch of 4chan memes in your product demos.