3 ms·

No, hallucinations are similar to but not quite the expected result which shows they are approximations. For example, you tell Dall-E to draw a picutre of a ma

by _8j50 4y ago

No, hallucinations are similar to but not quite the expected result which shows they are approximations. For example, you tell Dall-E to draw a picutre of a man smoking a pipe but then it draws the pipe coming out of the mans butt and instead of a head the man has a leg on his neck. This is approximation. Now if the pipe looked wrong or it was a trumpet or the guy's head looked weird but still a head maybe it knows what a head is and it just made a mistake and it knows that the pipe goes into the mouth so it will be somewhere near the face. Undesrstanding is a 3yr old child drawing terribly, approximarion is drawing really well but all wrong. All this applies to LLMs I used Dall-E because pictures are easier to talk about.

nl 4y ago
I haven't used dall-e much but I've never seen stable diffusion or midjourney make an error like that, unless of course deliberately promoted. You can see this because of the big deal people made about image gen tools getting hands wrong: it was the most significant error that was systematically occuring.
- _8j50 4y ago
  There us just something about the mistakes. Let me put it thid way, it nails down hands and faces great most of the time with stable diffusion for example, why would that ever be an issue at any point if it understood what hands and faces were? If I didn't understand what a hand was or how to draw it,I would never get it right. But if I do get it right quite a bit because I understand what the object is, then it makes no sense for me to have a significant error rate where hands and faces are deformed. An artist who knows how to draw hands and faces will never make that mistake, especially when self-correcting is so easy. The only explanation is that it is approximating based on what it learned from the large swath of training data. Kind of like if a human remembered answers to a multiplication table by memorizing every input,output and trend as opposed to knowing how to process the data within the rules of math and generate output. LLMs imo don't understand the rules of language and context they only approximate what a rule compliant system should output.
  - nl 4y ago
    As has been pointed out ad-nauseam, beginner human artists find hands very hard to draw too - and newer models aren't really making hand mistakes often. I've never seen significant or systematic errors with faces. It sounds a little bit like you haven't actually tried these tools. The kinds of errors you seem to think they make just aren't there in practice. I'd encourage you to try them out! > LLMs imo don't understand the rules of language and context they only approximate what a rule compliant system should output. This is an area I know a lot about. There are no real universal rules for English grammar. If you look at something like the Penn treebank you can see that English - as used by humans - is more exceptions than rules. The fact that LLMs outscore any rule based system merely means that our grammar rules are mostly things derived from how English is used in practice, not vice-versa.

by _8j50 4y ago

nl 4y ago

I haven't used dall-e much but I've never seen stable diffusion or midjourney make an error like that, unless of course deliberately promoted. You can see this because of the big deal people made about image gen tools getting hands wrong: it was the most significant error that was systematically occuring.

_8j50 4y ago
There us just something about the mistakes. Let me put it thid way, it nails down hands and faces great most of the time with stable diffusion for example, why would that ever be an issue at any point if it understood what hands and faces were? If I didn't understand what a hand was or how to draw it,I would never get it right. But if I do get it right quite a bit because I understand what the object is, then it makes no sense for me to have a significant error rate where hands and faces are deformed. An artist who knows how to draw hands and faces will never make that mistake, especially when self-correcting is so easy. The only explanation is that it is approximating based on what it learned from the large swath of training data. Kind of like if a human remembered answers to a multiplication table by memorizing every input,output and trend as opposed to knowing how to process the data within the rules of math and generate output. LLMs imo don't understand the rules of language and context they only approximate what a rule compliant system should output.
- nl 4y ago
  As has been pointed out ad-nauseam, beginner human artists find hands very hard to draw too - and newer models aren't really making hand mistakes often. I've never seen significant or systematic errors with faces. It sounds a little bit like you haven't actually tried these tools. The kinds of errors you seem to think they make just aren't there in practice. I'd encourage you to try them out! > LLMs imo don't understand the rules of language and context they only approximate what a rule compliant system should output. This is an area I know a lot about. There are no real universal rules for English grammar. If you look at something like the Penn treebank you can see that English - as used by humans - is more exceptions than rules. The fact that LLMs outscore any rule based system merely means that our grammar rules are mostly things derived from how English is used in practice, not vice-versa.