you will not be able to have a baby Einstein unless you are already a Nobel laureate...
patently false.
Neither of Einstein’s parents were Nobel laureates;
his father was a salesman, his mother was a pianist (among other talents).
you will not be able to have a baby Einstein unless you are already a Nobel laureate...
patently false.
Neither of Einstein’s parents were Nobel laureates;
his father was a salesman, his mother was a pianist (among other talents).
I’d like to make a simple suggestion for this and future posts. I note that you say you will continue to use ′ petertodd’ to distinguish from individuals named “Peter Todd”. The leading space, however, is elusive, can cause havok with line breaks, and is very very easy to miss. Might you instead consider _petertodd? Making the space “visible” with a symbolic underscore is much more visually direct, the italics are a well trodden quote replacement in bibliographies & references, and I think this tactic communicates the concept (of non-human identification, but rather token representation) more effectively. Good stuff, Matthew. Keep it up!
Why “set aside for now” the Japanese language tokens? I find those to be some of the most interesting, since my research indicates that all training datasets were supposedly pre-filtered (using Facebook opensource code) to have only english language in them. So how do 2 Japanese strings slip in?
Google Translate indicates that they are representative of the word “Thirty” and the phrase “Thirty-One Flavor”. Google Search brings up 1000s of pictures of young Japanese girls eating ice cream. My question is, how does one company’s marketing slogan make it into the top 100 glitchtokens when other far more market-savvy companies don’t?
So, imho, place “Baskin-Robbins + JP” alongside petertodd, SolidGoldMagikarp, and the rest of the GlitchPantheon.
Tasha McCauley seems very much like the X-factor in this conversation. I feel like the other 3 members motives can be fairly well implied by evidence / actions / publications. But McCauley has a very very low profile. John Loeber wrote an excellent timeline and analysis of the OpenAI board that I feel is useful for this conversation:
https://loeber.substack.com/p/a-timeline-of-the-openai-board
And I tried to piece together the salient data points of Tasha McCauley’s career & bio from all the oblique article mentions, just to collect them in one place:
https://gregoreite.com/who-is-tasha-mccauley-openai-board-member/
Enjoy.