Try OpenAI's GPT-3, it's so much more powerful than OpenAI ChatGPT and you can actually tell it things like "write me a X word essay", which ChatGPT refuses to do for example. Also allows you to tweak the repetitions, distinctiveness, etc., give it a try if you want
@@franciskafieh It's based off of it, but it's been optimised for conversations ("Chat"GPT), and GPT-3 was originally just for writing texts, and performs much better in the regard of just creating you a text. There's a reason why ChatGPT is free and GPT-3 has a limited free use period, since it uses much more resources on OpenAI's servers because of how much more powerful it is.
I will! Thanks @xenon! Lots more planned around AI and making use of these tools, their limits (…wait, there are limits?!) to create an interesting and productive life soon :D I’d definitely and specifically want to make use of Davinci and Codex (beta.openai.com/docs/models/content-filter) pretty soon to see what they can do, and I’d love to see the differences across OpenAI’s models - Reddit does say that ChatGPT is based on GPT-3.5 (so it is a more advanced model it seems? I don’t know how they actually compare), and just has more restrictions and is tailored to conversational flow - it will be interesting to compare them on tasks like these! Thanks for stopping by, and see you in the next videos! :D
@@VictorTan yes, chatgpt is very comparable to gpt-3.5 (interchangeable with gpt-3 in my comment.) but @xenon is right that the plain davinci model (accessible from the openai playground) may yield slightly better results, as chatgpt is optimized for chatting. in my experience, free usage limits from their UI (both chatgpt and playground) are comparable. Tl;DR: chatgpt is basically just the gpt-3(.5) davinci model but slightly more optimized for conversation rather than pure text generation
also, regarding their limits, ChatGPT is currently a "research demo" (aka they're probably going to pivot to a paid model in the future.) Sometimes, if you ask it too many questions, it will tell you that you've bypassed a certain request limit. Playground has similar (if the not the same) rate limits, but it might remain free forever as it is a space for developers to test prompts before using OpenAI's API (to generate responses for use in other apps.)
Yes you’re absolutely right @r0shan! Also, really importantly, it’s super fun to play with too! I think there’s a ton of hype about this though, with basically every third TH-camr and their second cousin twice removed freaking out that it’s gonna revolutionize the world, break the causality of the universe, and allow you to make a billion dollars copywriting for businesses *eyeroll* Jokes aside, the models truly are insane, and I can’t wait to test the limits of their capabilities on this channel; thanks for stopping by and see you in the next ones! :D
Try OpenAI's GPT-3, it's so much more powerful than OpenAI ChatGPT and you can actually tell it things like "write me a X word essay", which ChatGPT refuses to do for example. Also allows you to tweak the repetitions, distinctiveness, etc., give it a try if you want
tmk, chatgpt is built on top of the GPT-3 danvinci model. all they did was tune it a little to make it more "chat-like"
@@franciskafieh It's based off of it, but it's been optimised for conversations ("Chat"GPT), and GPT-3 was originally just for writing texts, and performs much better in the regard of just creating you a text.
There's a reason why ChatGPT is free and GPT-3 has a limited free use period, since it uses much more resources on OpenAI's servers because of how much more powerful it is.
I will! Thanks @xenon! Lots more planned around AI and making use of these tools, their limits (…wait, there are limits?!) to create an interesting and productive life soon :D
I’d definitely and specifically want to make use of Davinci and Codex (beta.openai.com/docs/models/content-filter) pretty soon to see what they can do, and I’d love to see the differences across OpenAI’s models - Reddit does say that ChatGPT is based on GPT-3.5 (so it is a more advanced model it seems? I don’t know how they actually compare), and just has more restrictions and is tailored to conversational flow - it will be interesting to compare them on tasks like these!
Thanks for stopping by, and see you in the next videos! :D
@@VictorTan yes, chatgpt is very comparable to gpt-3.5 (interchangeable with gpt-3 in my comment.) but @xenon is right that the plain davinci model (accessible from the openai playground) may yield slightly better results, as chatgpt is optimized for chatting. in my experience, free usage limits from their UI (both chatgpt and playground) are comparable.
Tl;DR: chatgpt is basically just the gpt-3(.5) davinci model but slightly more optimized for conversation rather than pure text generation
also, regarding their limits, ChatGPT is currently a "research demo" (aka they're probably going to pivot to a paid model in the future.) Sometimes, if you ask it too many questions, it will tell you that you've bypassed a certain request limit. Playground has similar (if the not the same) rate limits, but it might remain free forever as it is a space for developers to test prompts before using OpenAI's API (to generate responses for use in other apps.)
chatGPT is actually insane AI no wonder it has a lot of users lol
Yes you’re absolutely right @r0shan! Also, really importantly, it’s super fun to play with too!
I think there’s a ton of hype about this though, with basically every third TH-camr and their second cousin twice removed freaking out that it’s gonna revolutionize the world, break the causality of the universe, and allow you to make a billion dollars copywriting for businesses *eyeroll*
Jokes aside, the models truly are insane, and I can’t wait to test the limits of their capabilities on this channel; thanks for stopping by and see you in the next ones! :D
@@VictorTan HAHAH thats true