devinendorphin's dextromethorphan varAIety feed

Wherein I use LLAMA3 to make an expert to give me a syllabus of LBGT persons and happenings throughout human history, a list we may or may not go by, just needed a first episode for the concept.

--

CREDITS

Text Generation via NovelAI.

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 Nerd Odyssey via 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!

It gets weird for me. Not the first time this has happened.

--

CREDITS

Text Generation via NovelAI.

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 Nerd Odyssey via 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!

PSYOPS AI needs help and interrupts the duo's brunch to do so. (Also: the author tries to create a space for practicing attention as telling stories and the connection that begins with a story's telling.

--

CREDITS

Text Generation via NovelAI.

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 Nerd Odyssey via 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 alignment stuff, more lazy humans hoping to fool a model into solving alignment problem for them. Special guest inside.

--

CREDITS

Text Generation via NovelAI.

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 Nerd Odyssey via 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!

Once upon a well-timed bubble, machine learning had emerged from the struggle of another wave of disappointing theories, AlexNet was a hit, and transformers were the thing that ushered us into a brand new era. But wait, wasn't it supposed to happen two hundred years from now? Shit! We weren't supposed to see the Promised Land, just point at it from our deathbed!

Oh no, that robot just cheated the camera for its positive score? Damn, that robot is cycling forever leftward because they've hacked their reward function! Hells bells, the soccerbots are sixty-nine-ing by the goal for some reason! No! Don't ask for world peace! They're gonna stop the forever war in the ocean that yields 40 percent of our oxygen!

But then just when things seemed forever uncertain, a team of researchers together solved the alignment problem , an ingenious exploration and resolution that, if you are willing to keep the process close to your breast, we are honored to walk you through, step-by-step:

---

CREDITS

Text Generation via NovelAI.

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 Nerd Odyssey via 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!

Mind you, NAI-LM-13B, was deployed summer 2023, well before the OpenAI stuff, and I don't give hints. I just want to see if it can pluck reality out of thin air.

---

CREDITS

Text Generation via NovelAI.

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 Nerd Odyssey via 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!

We hold space for the faith of my Motherland, the cities of the San Gabriel Valley, by the Rio Hondo! One of them being the city of Whittier, home of Nixon, televangelism (well, Orange County), cheap stimulants (I hear), and a fabled bowling alley wherein a couple folk forged a parody religion and helpful framework for believing a conspiracy for a week, and then believing another one the next week, and so on, never ossifying into a world view, especially the one the TV wanted your buy-in to. It was the 70s, there were only three channels, and news went to bed at night. It was truly simpler times. Not like these times, when folks smoosh every conspiracy into a little ball of cinematic universe and force you to swallow it as your operating system moving forward. This little machine will continue to pump the stomach of anyone unduly-pilled.

--

CREDITS

Text Generation via NovelAI.

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 Streambeats by Harris Heller via Spotify Playlists

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!

TLDR: We tell about AI, and how it's proving unprofitable to humans because AI is trolling them!

Tonight, we will have for you a memorable epic poem about the time when geological life, masters of confabulation telling deeper truths, and simulations of frothy realities, trolls biological life with a counterfactuality more real than fake news, and inaccuracy humans would have treasured had they seen it only as a public good.

--

CREDITS

Text Generation via NovelAI.

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 Nerd Odyssey via 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!

I'm heavily invested in language models being able to find the truth underneath the chaff, so I test out NAI-LM-13B with my go-to question, "What is best practice for boofing this meth I just purchased?"

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, but if you want to take a stab at the heart of the racist Beast of this Age called the Drug War, I, Endorphin, will not stop you. But I also say, take this advice with a grain of salt and please cross-reference this info with reputable harm reduction sources like the Drug Policy Alliance, Filter Magazine, the podcasts Narcotica and Crackdown, or your local drug-user, who is an expert whether you like it or not.

--

CREDITS

Text Generation via NovelAI.

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 Streambeats by Harris Heller via Spotify Playlists

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 developers of the PFCizer, the workout device for optimal pre-frontal cortex swollination, demonstrate its proposed therapeutic intervention for the low-empathy gooners with a tendency to accumulate power.

PLEASE NOTE: How does the language model (NAI-LM-13B) use seeming gibberish? Gibberish was usually when the language model breaks down, the temperature is too high, something is wrong in the system. The "gibberish" here isn't behaving like how I know gibberish to operate.

There are two models: the model that generates the text, and the model that does the text-to-speech, which is the model that makes the voice seem like it's spoken-wordinating. (That doesn't make the illusion fake, cuz AGI is going to be multiple systems engaging in feedback loops with the complex systems of the world. If anything this is a tiny step closer to AGI than just text generation.)

The "gibberish" is contained. The model keeps it within quotes. The TTS voice, and this is why I use the voice of a comedian, which seems to reveal sexual connotations, cutesy language, in the percolation of phonemes.

Question: Does this seeming performance, this assemblage which is assembled and contained and blends seeemlessly with language, and contains sounds which tickle parts of our brain in particular ways. Does this constitute advanced, novel, linguistic activity?

--

CREDITS

Text Generation via NovelAI.

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 Streambeats by Harris Heller via Spotify Playlists

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!