🏅 Top traders

Resolves YES

As mentioned in an earlier update, it will resolve on the basis of this week's data (which includes 10/31), so not closed yet.

The set of people who watch Trevor Noah and who have the independence of mind to go google something is exactly zero:

Alternatively, fewer people watch him than ultra-niche podcasts

https://twitter.com/TheDailyShow/status/1585078690379345920?s=20&t=c8TQ1oIa_KZIPRJk-u41NQ

wonder how much this will impact things

RIP OpenAI—

You trained compute-inefficient, Orwellian-censored models so that others might spend their personal hedge fund money to build much better open source ones

Stable diffusion dipped below DALL-E this morning.

Two days left and DALLE isnt yet anywhere close

Though interestingly most of the traffic for Stable Diffusion seems to be coming from China

@JaimeSevilla one month, not two days

@jack oops

🤮

Dall-E now available without waitlist

https://beta.dreamstudio.ai/

Would not use this garbage if they paid me (per image)

Dalle finally goes “no waitlist” because no one wants it. Stable Diffusion traffic benefits more 🤔

@Gigacasting Non-believers getting wrecked rn (perhaps openai could release something that could spike it high for a day, i'm not paying close attention to if there are schedules or something, but there's no way stablediffusion is gonna slow down)

10 releases per day, or 1 virtue signal per month 🤔

Link: https://multimodal.art/news/1-week-of-stable-diffusion

Assuming this is, "stable diffusion line is above dall-e line" and not, "stable diffusion average interest over time is greater than dall-e's"

@GeorgeVii Right, this resolves based on only the point in time that includes October 31, not the average over time. Adding that clarification.

Probably the most important thing to happen in open-source AI since [Tensorflow/Word2Vec]

Long tail of improvements coming, and their main api also has no wait-list; “openAI” looking about as cool as a mainframe company with their approach that’s about as open or cool as a 1970s mainframe pc

when your engineers have 170 IQ and your product managers have no iq

@Gigacasting yes, I noticed that too. Extra uncomfortable since those images are of noticeably lower quality, probably because of less training data.

Seems like it could go either way pretty easily.