Posts

Comments

Comment by Rabrg on What do you do to deliberately practice? · 2022-06-05T17:12:15.311Z · LW · GW

Something I've recently started that I've found to be beneficial is deliberately practicing friendliness / extroversion / accepting "scary" social invitations that I previously would've rejected.

I've only been doing it for a couple of months, and I'm already seeing benefits like improved confidence, new friends / interesting conversations, and a deeper love for humanity.

Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-08T21:11:56.419Z · LW · GW

Hmm.. the second iteration of the second prompt isn't able to be pasted in that comment for some reason. Here it is:

Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-08T21:09:56.587Z · LW · GW

This is a great example of how even a single iteration on the prompt can vastly improve the results.

Here are the results when using your quotes exactly:

Pretty dreadful! But here they are, with the exact same prompt, except with ", digital art" appended to it:

Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-06T20:49:10.680Z · LW · GW

Unfortunately, the 2nd prompt violates the content policy (violence)

Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-06T20:46:39.568Z · LW · GW
Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-06T20:44:32.190Z · LW · GW
Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-06T20:14:56.679Z · LW · GW
Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-06T20:04:36.808Z · LW · GW

I could see your mother having a promising career in QA

Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-06T19:53:56.206Z · LW · GW

It did a fairly decent job, though it apparently prefers the colorized versions being on top.

Note that getting DALL-E to work well is a bit like GPT-3: you can usually get much better results by making a few iterations on the prompt.

Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-06T19:47:24.605Z · LW · GW

Cool idea. It seems that it isn't great at this, as only a few of the generations had exactly 7, but they are mostly all around there (6 or 8):

Comment by Rabrg on DALL·E 2 by OpenAI · 2022-04-06T19:30:13.295Z · LW · GW

I can post a few more non-cherry-picked generations here, if there are some captions / prompts anyone would like to try.

edit: alright, I should try to get at least a little bit of work done today. might come back to generate some more later

Comment by Rabrg on The Neuralink Monkey Demo · 2021-05-03T18:15:37.369Z · LW · GW

The hardware should be best-in-class due to the massive amount of channels (over 1,000), and the fact that each channel is surgically implanted into the head. For comparison, 16 channels is on the high end for consumer-grade BCI kits, and each channel is a sensor that rests on top of the skin.

As far as why they aren't making use of its capabilities to do something more impressive, I don't know. 

For what I would consider a more technically impressive presentation, see this video of a man controlling two prosthetics in 3d space to slice bread.

Comment by Rabrg on The Neuralink Monkey Demo · 2021-05-03T04:12:35.348Z · LW · GW

The hardware is impressive — it's best-in-class, but the presentation was mostly theatrics. We've had brain-computer interfaces for cursor control for 30 years (pong can be reduced to 1D cursor control — it's even simpler than the first task).

It's just a lot cooler to the public when its Musk getting a monkey to play a video game.