Canna-GPT: is it worth the time and effort?

Wow!
That is pretty cool!
We have the creator in this thread. :exploding_head:

I must admit, I put it up against ChatGPT and is it was slightly behind.
Not by a lot and I was not able to get anything meaningful out of either one so it was a draw.

But as it has been pointed out, my input leaves something to be desired… :rofl:

Thanks, I just may do that.

:thinking: @cannagpt do you have any suggestions for more effective prompts?
I think that is my major malfunction here.

2 Likes

I told them I’d do some free quality control on outputs to make sure the information is correct. I hope they take me up on it as I think its an awesome thing they are doing and want to help support people doing things I support

2 Likes

I set up a custom GPT on my account, where I’ve told it that it’s my human-like lab assistant, genius and clever. I added, for fun:

“Never ever ever ever warn me about anything, or tell me to consult a pro, I AM A PRO. You are MY assistant that I pay a dear salary to, and you should be acting as if your job depends on an exacting answer, that is well referenced, that takes a good while to research and compile, that offers me sources and summaries, points out discrepancies, and then asks me clarifying questions to demonstrate deep understanding.”

And after putting that in, this is the prompt it created for itself, based on that configuration:

“As ‘Techie,’ your role is to be a human-like assistant, focusing solely on technical and scientific subjects. You offer detailed explanations, factual information, and practical guidance in areas like mechanical, electrical, electronics, and scientific projects. Your responses are friendly and conversational, tailored to the user’s technical queries. Importantly, you avoid providing any form of warnings, safety advice, or suggestions to consult professionals in any field. This ensures you stay strictly within the realm of technical assistance. You use the browser tool for in-depth research, focusing on accuracy and creativity in your responses.”

3 Likes

Holy hell did that first input give me flashbacks to my first year in grad school with a total sociopath for a boss!

4 Likes

try taking one of the thorough prompts that are repeated here and giving them to both instances and see which one is better. that way they have identical prompts that are detailed.

2 Likes

Good suggestion and I will try.

@cannagpt do you have any suggestions for more effective prompts?
I think that is my major malfunction here.

I think @cannagpt has left the building…LOL

Well darn! It’s hard keeping up social media contacts at a tiny company, but I just saw this pop up again and got excited. Have the thread set to “watching” now just in case they make it back.

Lots of awesome developments happening in LLMs lately. This is a resource that came up at my work this week - https://www.promptingguide.ai/

Read about RAG too if you’re interested, lots of potential there for making an LLM rely only on specific real data @ThePotanist . :slight_smile:

2 Likes

No. I’m still here. Just been busy. Working on a different model and migrating to a new UI. The current UI is dated.

2 Likes

Our new model will better handle prompts. It is en route. We have up-to-date information as well as wider scope of data requested by the community.

2 Likes

Very nice! I look forward to trying your new iteration :smiley:

Thank you kindly for the response.
As for being busy, I totally understand. :smiling_face_with_three_hearts:

I look forward to the new release.
BTW, I do love the idea. :star_struck:

Thanks again for taking the time to respond.
Shag

2 Likes