5 ms·
A transformer is a universal approximator and there is no reason to believe it's not doing actual calculation. GPT-3.5+ can't do math that well, but it's not "j
by astrange 4y ago
A transformer is a universal approximator and there is no reason to believe it's not doing actual calculation. GPT-3.5+ can't do math that well, but it's not "just generating text", because its math errors aren't just regurgitating existing problems found in its training text.
It also isn't generating "the most likely response" - that's what original GPT-3 did, GPT-3.5 and up don't work that way. (They generate "the most likely response" /according to themselves/, but that's a tautology.)
- mach1ne 4y ago> It also isn't generating "the most likely response" - that's what original GPT-3 did, GPT-3.5 and up don't work that way. What changed?
- meow_mix 4y agoReinforcement learning w/ human feedback. What u guys are describing is the alignment problem
- mistymountains 4y agoThat’s just a supervised fine tuning method to skew outputs favorably. I’m working with it on biologics modeling using laboratory feedback, actually. The underlying inference structure is not changed.
- astrange 4y agoIt answers questions in a voice that isn't yours. The "most likely response" to text you wrote is: more text you wrote. Anytime the model provides an output you yourself wouldn't write, it isn't "the most likely response".
- afiori 4y agoI believe that ChatGPT works by inserting some ANSWER_TOKEN, that is a prompt like "Tell me about cats" would probably produce "Tell me about cats because I like them a lot", but the interface wraps you prompt like "QUESTOION_TOKENL:Tell me about cats ANSWER_TOKEN:"
- astrange 4y agoIt might, but I've used text-davinci-003 before this (https://platform.openai.com/playground https://platform.openai.com/playground) and it really just works with whatever you give it.
- afiori 4y agomeaning that it tends to continue your question?
- mort96 4y agotext-davinci-003 has no trouble working as a chat bot: https://i.imgur.com/lCUcdm9.png https://i.imgur.com/lCUcdm9.png (note that the poem lines it gave me should've been green, I don't know why they lost their highlight color)
- afiori 4y agoIt is interesting that the model seems unable to output the INPUT and OUTPUT tokens; I wonder if it learned behavior or an architectural constraint
- mort96 4y agoYeah, that's an interesting question I didn't consider actually. Why doesn't it just keep going? Why doesn't it generate an 'INPUT:' line? It's certainly not that those tokens are hard coded. I tried a completely different format and with no prior instruction, and it works: https://i.imgur.com/ZIDb4vM.png https://i.imgur.com/ZIDb4vM.png (again, highlighting is broken. The LLM generated all the text after 'Alice:' for all lines except for the first one.)
- afiori 4y agoThen I guess that it is learned behavior. It recognizes the shape of a conversation and it knows where it is supposed to stop. It would be interesting to stretch this model, like asking it to continue a conversation between 4-5 people where the speaking order is not regular and the user is 2 people and the model is 3
- ainiriand 4y agoI wonder if that was why when I asked v3.5 to generate a number with 255 failed all the time, but v4 does it correctly. By the way, do not even try with Bing.