[LooerCell]

Creator here, just woken up and it’s funny to see PizzaGPT on the front page of HN.

I had the idea yesterday, when reading about ChatGPT being blocked in Italy. I’m an Italian living abroad so I still have access to it, but my parents and friends living there don’t. I believe AI is a revolutionary tool that should be available to anyone, the same way internet has been.

The website was coded yesterday in a couple of hours, so it’s full of bugs and potential improvements. But I just wanted to ship something quickly. For the stack I used Nuxt 3, Tailwind and DaisyUI.

Regarding the AI, it’s literally just a wrapper of OpenAI completions API (turbo-3.5 model) and a chat interface, so it should give the same answers as the free version of ChatGPT.

I am still a bit afraid of the costs if many people start using it, so I’ve added a donate button. You can donate the equivalent of a pizza (pizzaware model?) to help keep it running

[Al-Khwarizmi]

> Regarding the AI, it’s literally just a wrapper of OpenAI completions API (turbo-3.5 model) and a chat interface, so it should give the same answers as the free version of ChatGPT.

Have you tested this? My impression after a couple of tests is that its answers are worse than free ChatGPT (no idea why).

Maybe I just got unlucky with the RNG, though.

[dmarchand90]

The free user interface does something behind the scenes to provide context to chatgpt but I haven't figured out what it is

[blatant303]

> My name is George

Nice to meet you, George. How can I assist you today?

> What is my name

As an AI language model, I don't have access to your personal information such as your name. Can you please tell me your name?

You need to pass the whole chat history in your request:

    [{:role "assistant", :content "Ciao, sono PizzaGPT come posso aiutarti?"}
     {:role "user",      :content "My name is George"}
     {:role "assistant", :content "Nice to meet you, George. How can I assist you today?"}
     {:role "user",      :content "What is my name"}
     {:role "assistant", :content "Your name is George."}]

On the ChatCompletion API endpoint

[dmarchand90]

I think it is doing something a bit sophisticated in choosing what to pass in the context when the conversation grows beyond the token limit

[blatant303]

https://python.langchain.com/en/latest/modules/memory/types/...

Maybe ?

[LooerCell]

Yes, I know, but it was a deliberate choice not to send the entire history, to keep the token consumption and costs low

[LooerCell]

Not really had much time to test the results. Yeah it should be possible to tweak the settings (mostly temperature and prompt) to get better results. Also I think I’m not sending the full history right now, to avoid consuming too many tokens

[kgeist]

>Also I think I’m not sending the full history right now, to avoid consuming too many tokens

You can ask it to summarize the dialog, thus compressing the dialog history into a few sentences. That will give you some basic dialog context.

[eric-burel]

Does it address the issue that led to forbidding ChatGPT in the first place ?

[crakenzak]

no.

[justaj]

Feedback:

The prompt bar seems to be in front / on top of the (generated) text. Pretty annoying.

[LooerCell]

Thanks for the feedback! Tried to solve this just now, can you check?

[justaj]

Yep, it seems to be much better now. Thanks for the fast response!

[alignItems]

I think if you want the same as ChatGPT you need the use the chat API, not the completions API.

[LooerCell]

It's confusing because is actually called Chat Completion API :) https://platform.openai.com/docs/guides/chat It's the one used on ChatGPT that includes gpt-3.5-turbo and gpt-4