Not yet, but coming up next week!

hey jessica! while agents can perform rag, i wouldn't say that crewai replaces regular rag applications. agents operate at a higher level of complexity than regular rag apps. and i am an advocate for the simplest possible solution that delivers the job. that being said. there is one thing called "agentic rag" which can enhance the results of a rag application with an agent that oversees the retrieval process. however, i don't think you would necessarily use crewai for that one!

it's my pleasure!

i appreciate it!

thanks! i will :)

@@alejandro_ao my class is over just now and we learned decision tree ... Please upload all models videos

where? do you mean i made a typo and the llm still understood?

@@alejandro_ao At 37:30 roughly.The observation "1.6425e+24" seems totally wrong, how did you get this number (you pasted that in manually). Then, the model still answered with the right (but different to your observation) answer "2.98625e+25". This means (if i got that right) the model ignored your observation entirely and ended up doing the math itself. Did I miss something? While it is impressive that the model realized this number is wrong it makes me wonder how predictable a model can follow orders. It seems there is no guarantee at all and it's so easy to manipulate also. Imagine you have a website offering you some kind of services. Through prompt injection you can infiltrate an actor with access to agents (and potentially secrets) and rewrite the prompt to do what ever you want. I see almost no real use case besides demos as it is so inherently impossible to secure against misuse. LLMs are fascinating search engines through data but they are almost impossible to control safely either.

i appreciate it :)

hey there! they offer a hosted service where they turn your crews into apis, but yeah that is paid. howerver, as far as i understand, it comes with much more than that as they actually personally help you set up your workflow with their experts

wonderful time to be alive indeed

thank you!! I sure will!

🚀

hey there! if you want to deploy an app for many users there is sadly no free solution as even the free tiers of most llm providers have a rate limit. but if you want to deploy this for a small team, say 20 people, you can probably use Groq models! they have a free tier with a limit on the number of calls you can do per minute

it's basically a regular LLM, but that is especially good at calling functions. you could use it for general purposes too (like gpt-4 or llama3,) but it would be worse than them on most general tasks. that's why they recommend that you implement a routing layer where if your request involves calling a function , it uses the groq model, and for everything else, it uses the general model

Use sentence-transformer==2.2.2

I would recommend that you only do this after thorough testing. and very importantly, do not give write privileges to the MySQL user that you are using to access the database. You don’t want your LLM to be able to update or delete data in your database 👍