@retoor

https://archive.is/gRIXH

comes up 404.

what I thought. Uninterested in being doxxed.

@retoor "click click click"

no no no.

@retoor I normally don't click external links on devrant and I have link preloading disabled also. Too many people trying to do others lately, and it drives the left-right divide as a result.

You appear to be someone important, or at least very skilled. Spook shit is fun.

@cprn Always approach an armed man cautiously. And for unarmed men always offer to assist them with their wheelchair.

I go for hikes with NVGs.

Alone.

In the woods.

At night.

Instead of training the networks, just instance the graph multiple times into several variations, and select subsets of weights and biases from each instance. Randomly adjust or reinitialize those subsets, then do a few minibatches and measure the MLE of each instance and merge the most accurate one (for some flavor of 'accurate') back to the parent graph.

No backprop needed in training.

I'm trying to find a variation of this that works for forward passes during inference, but I haven't yet.

Hypothetically this should work better than even AdamW because a sufficiently random source should always approximate the gaussian distribution for a given training regime way better than any amount of training and validation data you can get your hands on.

No local minima that can't be overcome!

People like me and a hundred thousand others will just stop giving the US and affiliate companies/governments our ideas and opensource contributions and develop in private instead.

The governments of the world will lose the public-private armsrace in AI as a result.

This is a great development. It means the governments, the real people in charge, are fucking stupid as rocks.

@retoor the stationary distribution has converged to the steady state!

Almost as unpredictable as setting the nuclear launch codes to straight zeroes.

They'll never suspect the password is password!

@typosaurus you know what they say, no hole like a blowhole!

If I'm ever a fabulously wealthy silicon valley yuppie, you're officially invited to the orgies.

@iSwimInTheC The space cowboys are lacking sufficient yeehaw I see.

@Demolishun I have risen from my crypt. The magic eightball says "there is fortune in the sunset."

It's an eightball. Eight is a number. Thats math!

@retoor First this is very cool.

Second, and this is so stupid I used to do it regularly, but apparently var = [] is not the same thing as var = list()

No idea if that has any application to your memory leak, but I recall having a couple memory leaks because of that specific issue.

You're building in python, what libraries you using?

@retoor I have on the order of a couple hundred books. A respectable collection.

If I had a scanner I could OCR that shit.

If I had the time I could build the scanner.

Alas.

@bazmd what kind of cooking? Is that you walter?

@figoore about tree-fiddy.

We need a new devrant.

With private messaging.

I'll add in the obligatory "with black jack and hookers!"

So we all don't lose the connections, ephemeral or otherwise, that we made here.

But would they have actually aborted?

Imagine being those astronauts, all that training, so close to something they trained years for, knowing they would likely never get a second chance.

I think they might have ignored the errors and warnings, and gone for it anyway.

255 days later and not a single soul responded before now.

But it was a good joke fwiw.

I should post my voice so you can have bots that say malicious things, but with a real voice.

@Demolishun But what if skynet kills me first bro?

@atheist Regression trees with very large inputs in X.

Say with numbers that have 400-500+ digits.

@jestdotty who doesn't want to build cool shit?

But no, its just what people are interested in.

@retoor apparently you were right that I only visit once ever four days.

The bread thing is someones epic idea of a prank.

I guess they better get to making that bread!

@Demolishun your aunt was totally fucking wild by the way.

@jestdotty my process and experience as well.

Don't know what reduced your functioning but I'm sorry to hear it. What a nightmare.

I think most people just enjoying reading cool shit even if I don't explain all of it. Technical details tend to bog shit down and the big picture gets lost in the minutiae. You can even see interest drop off the longer a post is, so usually I try to keep them at a couple pages max.

Sometimes other Ranters will come in with "thats not possible" or "thats stupid", and either I realize it is, or I get to do the fun thing and say "but I just did."

Btw, what kind of sickness was it? (if thats not prying too much)

@figoore Now you're just fucking with me!

Be well figoore, the future is bright for you and everyone on devrant....

Until one of us builds the killer robots.

@Demolishun henceforth you will be our resident expert in all things cult and occult.

@figoore It is always in the room. Always by our side, like a companion that never leaves.

@jestdotty you stumble on shit thats already been discovered too?

257 days since your last post.

What, you been busy or something?

Too busy for all of us on devrant?

We're your real family.

Drink the koolaid.

ONE OF US. ONE OF US.

@Demolishun the very friction you complain about is the thing that motivated the automation in the first place.

The game is not the SaaS product. The game is a never ending loop of searching for and finding SaaS products that reduce friction further.