i love this! also super cool that you also dig hf! i’m curious about where we are all going next in machine learning… researchers are discovering the limits of LLMs because they are finding lags in translating the math into the language/word tokens and back n forth. VLMs are a new thing i’m seeing in the space recently to assist in translating our words for images == Visual Language Models. I am seeing the ways researchers are taking each of the 5 Human Senses, and discovering how to ‘translate’ that into computer code / math stuff. ty4s! stoked to hear your thoughts on it!
in a way, doesn’t every component of mathematics have a corresponding word? i.e., one and two for 1 and 2? it might be a bit of a chicken or the egg thing but we had to speak math before we ever wrote it down in its own symbols, no?
one of the things i find so amazing about natural language is its capability for infinite expression.
haha i like the chicken/egg analogy! Agreed we speak math... i think because the computers work on a binary system of 1/0, we have to structure the information it is processing down to numbers for it to work as quickly as possible. i do love the infinite expression we create as humans... i guess the question is... can we teach a computer to 'create' the same way we do?
I remember citing the "black box" idea of modern LLMs (marketed as read: AI) while talking to a friend of mine who did his Master's in applied linguistics and went on to have a career in conversation design. Only for that to somewhat be swept away in favor of what is now the "AI bubble." Anyway, he was very annoyed with me saying that we DO know how they work! I don't remember his explanation, I just remember his incredulous response when I cited the oft-mentioned phrasing 😅
loving this topic and stoked to see the hf mention lol
feel whole heartedly huggingface is the best example of what ai can be. they don't get nearly enough credit for what they've achieved.
maybe it's the english major inside me, but it seems more and more to me that language itself is the stuff of consciousness.
the inner monologue.
the large language model seems to exemplify this.
i am not my body or my eyes.
i am the language i think and speak, or rather language speaks through me.
i love this! also super cool that you also dig hf! i’m curious about where we are all going next in machine learning… researchers are discovering the limits of LLMs because they are finding lags in translating the math into the language/word tokens and back n forth. VLMs are a new thing i’m seeing in the space recently to assist in translating our words for images == Visual Language Models. I am seeing the ways researchers are taking each of the 5 Human Senses, and discovering how to ‘translate’ that into computer code / math stuff. ty4s! stoked to hear your thoughts on it!
that’s intriguing i hadn’t heard about this math limitation. as an english major math always just seemed like an extension of language
agreed! math really has it's own way of being a language itself! so things can get lost in translation
in a bit of serendipity this also just came across my feed. in classic open ai fashion they’ve no plans to release this model lol
https://www.lesswrong.com/posts/RcBqeJ8GHM2LygQK3/openai-claims-imo-gold-medal
in a way, doesn’t every component of mathematics have a corresponding word? i.e., one and two for 1 and 2? it might be a bit of a chicken or the egg thing but we had to speak math before we ever wrote it down in its own symbols, no?
one of the things i find so amazing about natural language is its capability for infinite expression.
haha i like the chicken/egg analogy! Agreed we speak math... i think because the computers work on a binary system of 1/0, we have to structure the information it is processing down to numbers for it to work as quickly as possible. i do love the infinite expression we create as humans... i guess the question is... can we teach a computer to 'create' the same way we do?
I remember citing the "black box" idea of modern LLMs (marketed as read: AI) while talking to a friend of mine who did his Master's in applied linguistics and went on to have a career in conversation design. Only for that to somewhat be swept away in favor of what is now the "AI bubble." Anyway, he was very annoyed with me saying that we DO know how they work! I don't remember his explanation, I just remember his incredulous response when I cited the oft-mentioned phrasing 😅