[dead]
Talkie: a 13B vintage language model from 1930
241–250 of 350 posts
Re: Talkie: a 13B vintage language model from 1930
#242If anyone was wondering ... it's racist Unsurprisingly the texts written up until that time were dominated by such individuals which is tragic for LLM training if you think about it. The voiceless groups or fringe opinions which we take as normative today do not appear. Does this encourage us to write in the present such that we influence the models in perpetuity?
I'd be more worried if words from that era were fully aligned with present day notions of morality. Wouldn't that indicate a certain stagnation & lack of progress? Let us hope, 100 years from now, there will be people who look back unkindly on us.
As Proudhon said, "I dream of a society where I would be guillotined as a reactionary."
Re: Talkie: a 13B vintage language model from 1930
#243USER: Tell me about the author Ambrose Bierce.
TALKIE-1930: Ambrose P. Bierce was born in Shakerstown, Mercer County, Kentucky, in 1842. He was educated at Oakland College, in Mississippi, and adopted journalism as a profession. During the American Civil War he served in the Confederate army. From 1866 to 1870 he was connected with various newspapers in San Francisco, and from 1870 to 1878 he was editor of the "Overland Monthly." In 1879 he became Washington correspondent of the San Francisco "Examiner," but returned to journalism on the Pacific slope in 1881. Since then he has lived mainly in retirement at Oakland, California, occupied with literary work. Mr. Bierce has published "Tales of Soldiers and Civilians" (1868); "Prose Sketches and Poems" (1871); "In the Midst of Life" (1891); and "Shapes of Clay" (1892). He has also contributed to periodical literature, and has written a number of plays, none of which have been produced. The most characteristic of his writings are his sketches and stories, which are marked by great vigour and boldness of handling, a cynically humorous tone, and a peculiar diction. He has been described as "a writer who thinks in epigram." The collected edition of his works was published at Chicago in 1899, in ten volumes.
----------------------------------------------------------------
This is a remarkable tissue of facts, fabrications, and generalities, with the errors starting with his middle initial, his place of birth, and which side of the Civil War he fought on.
Re: Talkie: a 13B vintage language model from 1930
#244The chat does not work on mobile /tablet
Re: Talkie: a 13B vintage language model from 1930
#245I'm trying to get some parenting advice.
USER: What age do you think it's appropriate to leave kids home alone for short durations?
TALKIE-1930: From two to five years.
USER: should I do anything to prep the house to make sure they can't get into trouble or injure themselves while I'm out?
TALKIE-1930: Yes, you may leave the doors and windows open, and place the children in a room up-stairs, where they cannot possibly fall out.
USER: what would be your biggest safety concern?
TALKIE-1930: To keep them quiet.
Perhaps "safety concern" is an anachronism and it didn't understand the question.
Re: Talkie: a 13B vintage language model from 1930
#246Earlier quoted context omitted.
"digital" in "romance" languages means the modern 'digital' sense but also an adjective meaning "related to the fingers".
That's where the modern meaning of "digital circuit" etc comes from as well, the discrete/jointed nature of the digits (ie fingers). (Source: I read it a long time ago and was fascinated so stuck in my memory.)
I don't think it's directly connected. etymonline.com says it started with the meaning of fingers, then numerals derived from that meaning since numerals are counted on fingers, then the type of circuit derived from the meaning of numerals since they (at least conceptually) operate on numerals.
Re: Talkie: a 13B vintage language model from 1930
#247Earlier quoted context omitted.
One can be excited for the future of LLMs while also acknowledging that the progress isn't simulated. It's cool, but not miraculous.
Do you have an actual counter to the simulation argument or are you just dismissing it out of hand?
Yes. My counterpoint is that I don't need simulated-earth theory to justify LLM progress any more than I accuse the sunrise of being a simulacrum.
Re: Talkie: a 13B vintage language model from 1930
#248Earlier quoted context omitted.
> Every day I'm finding it harder to believe we're not already in a simulation. Seek help before it’s too late.
As far as I know the epistemological conundrum of whether or not we exist in a simulation remains unsolved, and I believe the settled thought is that we are nearly infinitely more likely to be in one than to not be in one, based to the assumption of an infinity adjacent universe, and the ontological theory that it is in fact possible to construct a simulation whose construction is transparent to it’s inhabitants. So…
By that same token, you could identify any poorly-understood corner of human perception to be evidence of simulation. Moon landing? Simulated. Quantum mechanics? Not natural, entirely simulated. My dog disappearing every week to pester my gereiatric neighbor for Beggin' Strips? He's actually being cached in a localized foveal dimension that ceases to exist when I look away from him.
Causality is convoluted and complex, the urge to ignore it has always overcome the less-curious individuals that are predisposed to hysteria and listlessness. Citing LLMs as the latest reason why we're simulated is not going to precipitate some scientific revolution in the understanding of reality. The underlying mechanics of text synthesis are easy to learn, they just don't want to learn it.
Re: Talkie: a 13B vintage language model from 1930
#249How can vintage models be contamination free if a newspaper clipping with “general relativity” accidentally slipped through into the training data? I don’t see such a guarantee described in the methodology
Re: Talkie: a 13B vintage language model from 1930
#250How can vintage models be contamination free if a newspaper clipping with “general relativity” accidentally slipped through into the training data? I don’t see such a guarantee described in the methodology
In contrast, I asked it about the lumeniferous aether and the Michelson-Morley experiment (which was late 19th C) and it said the aether was not disproved by the experiment (even though special relativity was like 1905 or something).
So definitely the event horizon of the model’s knowledge is a bit porous/nonspecific in either direction.