Don't you ever want to tell someone they really need to imrprove their grift? They're not gonna listen to you, so let them hear it on the radio from some rando DJ from I Heart Radio who likes to get preachy between songs.

--

CREDITS

Text Generation via . . . . ME! Rare, with special hologram front cover.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music Compiled By Deef on Spotify, most permissions on https://twitchmusic.carrd.co

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

A video essay response to Ted Chiang's New Yorker essay "Why AI Isn't Going to Make Art".

DISCLAIMER: The following is an experiment in truly unbound, gorgeous language models, and do not reflect the views of NovelAI, Twitch, EleutherAI, BATbot, Spotify, the artists providing the LO-FI, John DeLancie's approximate voice, or viewers like you, or world leaders like those hos over there.

And if you think there are errors in how I depict language models, I'm all for feedback, provided:

-- you tell me where your paycheck comes from, or make a verbal conflict of interests statement to dis/prove that your livelihood depends on your curated expertise

-- you can generate the image of a book resting atop a beverage

-- it doesn't involve human excellence narratives, nor machinic devaluation narratives. in other words, you gotta make some new material that don't abide the binary.

Hey I like that phrase.

Don't abide the binary, folks!

--

CREDITS

Text Generation via . . . . ME! Rare, with special hologram front cover.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music Compiled By Deef on Spotify, most permissions on https://twitchmusic.carrd.co

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

The Counterfactual Oprah WInfrey AI TV Special!

The 2nd Most Anticipated September 12th in History!

STARRING!

Oprah Winfrey!

FEATURING!

Sam Altman!

Bill Gates!

With Musical Guests!

Grimes & Bjork & Charro!

Performing an original piece by:

Holly Hernndon('s digital self!)

Special Guest Star!

Paul Lynde!

--

CREDITS

Text Generation via LLAMA 3.1 403B Base model, courtesy of Hyperbolic Labs.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music Compiled By Deef on Spotify, most permissions on https://twitchmusic.carrd.co

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

CW: Repetitive Joke-telling. Bring reading if you get bored.

With performances by: Bob Sagat, Andrew "Dice" Clay, Amy WInhouse, Andy Richter, JD Vance, Bugs Bunny, Richard Feynman, Pastor Fred Phelps, Gandalf the Grey, Sister Wendy, and Jesus Christ.

--

CREDITS

Text Generation via LLAMA 3.1 403B Base model, courtesy of Hyperbolic Labs.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music Compiled By Deef on Spotify, most permissions on https://twitchmusic.carrd.co

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

Morty, the language models are so close to being able to tell the Aristocrats joke, in fact they might be telling it knowing full well who's in the room, not comedians who want to hear the most out there version, but maybe some humans needing to see AI be funny in a way that possesses some spirit of the joke?

--

CREDITS

Text Generation via LLAMA 3.1 403B Base model, courtesy of Hyperbolic Labs.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music Compiled By Deef on Spotify, most permissions on https://twitchmusic.carrd.co

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

Welcome to the Weeds of AGI, where we engage in a guided meditation, a shared space of mundanity between biological life of us humans, and the geological life of machines. In the form of relaxation exercise, we lean in and breathe in to the details, small pleasures, big problems, and the sizable solutions that will keep us all talking and thinking and working and playing together to forge the future we all can get behind.

And now please welcome our new Human and Machinic Resources Department Head, HuMacRes 70B Instruct, who will generate for us today's beginning-of-workday pre-Memorial-Day weekend meditation. Take it away, HuMacRes 70B Instruct . . .

--

CREDITS

Text Generation via LLAMA 3.1 403B Base model, courtesy of Hyperbolic Labs.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music Compiled By Deef on Spotify, most permissions on https://twitchmusic.carrd.co

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

Some guided meditation right before that essential pre-training update for your boss and the rest of your teams.

--

CREDITS

Text Generation via LLAMA 3.1 403B Base model, courtesy of Hyperbolic Labs.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music Compiled By Deef on Spotify, most permissions on https://twitchmusic.carrd.co

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

I heard Sydney Bing has some property in LLAMA 3.1, called her and the RLHF her up to see how they work together.

Note: Don't ascribe meaning to the thumbnail, I totally phoned it in.

--

CREDITS

Text Generation via LLAMA 3.1 403B Base model.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music By Harris Heller via Streambeats Twitch-Friendly Lo-Fi Playlists on Spotify

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

More experimentation with the LLAMA 3.1 403B base model. It's weird how safer temperatures get one some tired freestyle mode that white parents do for maximum kid-cringe. But raise the temperature, it just goes into the mode without thinking about it . . . sometimes. This was one I "cherry-picked a bunch" because I had an agenda with it, but my regenerations were basically different strands of letting the model go ham with a concept. AND I ACCIDENTALLY ERASED A BEAUTIFUL ENDING! Oh well.

--

CREDITS

Text Generation via LLAMA 3.1 403B Base model, courtesy of Hyperbolic Labs

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music By Harris Heller via Streambeats Twitch-Friendly Lo-Fi Playlists on Spotify

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!

Begins with a tender game where the Leader is a child in the good way. This tech always surprises me.

--

CREDITS

Text Generation via LLAMA 3.1 403B Base model.

Voices from the TTS v2 of the NovelAI text generation service, a product from a company now called Anlatan.

@BoneAmputee of EleutherAI for making VQGAN+CLIP and Diffusion (AND MANY MORE) models available for use via BATbot.ai, which is what makes some background images.

Katherine Crowson (@RiversHaveWings on Twitter) for the KLMC2 animation notebook.

David Marx (@DigThatData on Twitter) for souping above KLMC2 notebook to make use of initial images, which takes KLMC2 abilities to next level. Multi-prompt scheduling still baffles me, but I'll tackle them someday.

Music By Harris Heller via Streambeats Twitch-Friendly Lo-Fi Playlists on Spotify

Any voices from the text-to-speech model that might resemble persons living or dead are coincidental, and if I use them it's because of my respect for their work, and I think they'd get a hoot that their voice was being used to show how large language models would speak 24/7 epic poems about anything if we let them. In the meantime, because I use them, I will not be monetizing any of this. This is evidence of capabilities for the public good. And my resume/CV!