Boxhead @NBoxhead

I assume this will be later used as evidence against me. theboxhead.itch.io Joined June 2021

Tweets

110
Followers

50
Following

291
Likes

186

Boxhead @NBoxhead

2 years ago

LLM's are predictive models for language... but are languages predictive models for humans?

0 0 0 20 0

View Details

Boxhead @NBoxhead

2 years ago

@rekdt There’s a very good reason that forbidden knowledge is acquired through profane offerings to beings that are terribly bothered by being roused.

0 0 0 30 0

View Details

@Devon_Eriksen_ Too much junk code. Shortcut the argument to the objective standard of “Consciousness”. There isn't one. Its why I like the Latin root per wiki: 'conscius' meabing “knowing with” or “having joint or common knowledge with another” Someone new to "know with"? Yes, please.

0 0 0 22 0

View Details

Boxhead @NBoxhead

3 years ago

Happy birthday buddy.

0 0 0 38 0

View Details

Boxhead @NBoxhead

3 years ago

@nickfloats @runwayml @ideogram_ai Invert the indicators: In example 2, 100 at a 500 valuation would indicate that they’re buying a ~20% stake. It only show the value that the purchasers perceive the total value to be… not actual value or market value.

0 0 2 657 0

View Details

Boxhead @NBoxhead

3 years ago

@AnthonyDeSimone @NoContextHumans Sunk-cost fallacy. C: Cash D: Theft-expense The source of your customers funds are unrelated to revenue recognition.

1 0 1 246 0

View Details

ひさだん @hisadan

3 years ago

//#つぶやきProcessing int n=999999,p[]=new int[n],i,j; float t=1; void setup(){ size(800,800); for(i=2;i<n;i++)if(p[i]==0)for(j=i+i;j<n;j+=i)p[j]=i; } void draw(){ clear(); stroke(-1); for(i=2;i<n;i++)if(p[i]==0)circle(i*sin(i*t)/99+400,i*cos(i*t)/99+400,2); t+=1e-7; }

28 319 2K 160K 502

View Details

Boxhead @NBoxhead

3 years ago

@wanyeburkett @NickADobos It’s how an engineer sees a problem. The connect is even needs higher up the hierarchy can be a factor. Destructive tendencies as malnutrition is to the need for food.

0 0 0 58 0

View Details

Boxhead @NBoxhead

3 years ago

@NickADobos Wanna get even lazier? tell Cursor to "wrap in bash heredoc", and you can implement code with the 'Run In Terminal' button. You can even specify a directory with the with an '@' reference.

1 0 1 379 2

View Details

Boxhead @NBoxhead

3 years ago

We never asked if AI *should* be taught to ride a Gecko. In our hubris, we only asked if it *could*.

0 0 1 51 0

View Details

Boxhead @NBoxhead

3 years ago

This took way longer than expected... Reworked how Agents control "walkers". Now they can 'Focus' on an object rather than constantly scanning and inferring from there. Also added some snazzy IK. #Unity #gamedevelopment #AI

0 0 0 84 0

View Details

Boxhead @NBoxhead

3 years ago

CGPT:CI is amazing and all... but it's blatant honesty and ability to show how messy I've let things get while 'Just Quickly Trying This', is quite frankly rude. It's like having someone working over my shoulder delivering shame just as I ask for it.

0 0 0 48 0

View Details

Boxhead @NBoxhead

3 years ago

Trying to get the general look right. Fell in love with a design that I saw elsewhere, but hoping it isn't dead. Hoping to go somewhere similar with the look.

0 0 0 47 0

View Details

Boxhead @NBoxhead

3 years ago

It's almost like teaching something to walk feels like it should take YEARS #unity3d #unity #animation

0 0 0 66 0

View Details

Boxhead @NBoxhead

3 years ago

Playing around with different representations of the "Brain" behind SISABUS. TL;DR - Stuff goes in on the left, Stuff happens in the middle, Stuff comes out on the right. #unity #mlagents #MachineLearning #BrainStructure

0 0 0 125 0

View Details

Boxhead @NBoxhead

3 years ago

Well that's not right... #Unity #MLAgents #bug

0 0 0 76 0

View Details

Boxhead @NBoxhead

3 years ago

13 Years of SpaceX Florida Launches in Under 1 Minute youtube.com/shorts/P-na5q6… via @YouTube

0 0 0 34 0

View Details

Boxhead @NBoxhead

3 years ago

I think people should say "Please" & "Thank-you" when using #ChatGPT & other imitative systems.

0 0 0 49 0

View Details

Boxhead @NBoxhead

3 years ago

#Doink

0 0 1 43 0

View Details

Boxhead @NBoxhead

3 years ago

Followed up! Most of the work happens in the first million steps, everything after that is fine tuning it seems...

0 0 0 41 0

View Details

Boxhead @NBoxhead

3 years ago

More graphs that are mostly meaningless foe anyone but me, but this is the VM training compared against local training. At this level of training it looks like CPU exclusive doesn't make too much of a difference. #MachineLearning #tensorboard