TOOL LLM for API calls explained

preview_player
Показать описание

Goal is to achieve a similar API performance by smaller LLMs as compared to ChatGPT's tools performance.

Arxiv pre-print (all rights with authors):

#aieducation
#gpt4
#chatgptprompts
Рекомендации по теме
Комментарии
Автор

I will come back to this video in 5 years to show my kids what google looked like before AI killed it

verasalem
Автор

May be slightly misleading if you dont consider the size of these models. Biggest Llama is 65B while smallest ChatGPT (3.5-turbo) is 154B!

shashankpathak
Автор

nice research!!!
I am always a fan of you.

chenkiwi
Автор

Thanks for this nice break-down. I'm curious to know: do they do that only for clickbait or just ignorance/laziness? It doesn't favor the LLM. Paradoxically: Your video made me want to test the LLM more than their videos probably would. Knowing that it is probably close to GPT-3.5 is a better selling statement to me than "mmm... this beats GPT4", "this changes everything", etc, etc.

MrMiguelChaves
Автор

Thanks once again… you keep things honest 😂

arundona