Show HN: I Turned DOOM into a Transformer
github.comInstead of trying to run DOOM on things, I have instead decided to use DOOM (many, many, many doom games actually) to create a transformer. For a lot of the math required I essentially had to use DOOM game mechanics and voodoo dolls (DOOM is turing complete btw) and exchange them with the actual operations, essentially creating an alternative to NumPy (a 65000x slower one). Thus, I am proud to present DOOMFORMER. Here are a list of things DOOMFORMER can do :
It can classify MNIST digits, even if you write these digits on the map!
It can talk(kinda)
You can use it to make decisions, so, it can play DOOM. Yes! You can play DOOM with thousands of DOOM games now! The marine's mind can be controlled by DOOM, playing on a map generated by DOOMFORMER. Truly, the possibilities are endless, and even larger-scale training of DOOMFORMER could be done on a measly T4.
If anyone would like to help me scale this up and make crazier things, I'd love to work together.
Was there even a human involved at all, or was this a "no mistakes" prompt outcome?
Nothing on this page looks made by a human, its _all_ generated, not a tiny bit of it, _all_ of it.
Guess having the word "doom" and "transformer"in your publication is enough today.
Awesome
>a handwritten digit takes 163 seconds of real-time DOOM in one map at the default schedule and 8 seconds at --opt 3.
What does this refer to? What does schedule mean here, and what are these opt levels?
Edit: oh sweet Jesus
https://github.com/mottopanikeiku/doomformer/blob/main/docs/...
Oh, it's AI word-vomit.
I pity the AI agent that had to work on this
Does that mean the title should have been "AI Turned DOOM into a Transformer"?
It helped me write my docs!
theres no pleasing you people do you expect me to write all my docs and disclaimers by hand
You won't like the answer to that question ;)
Yes. Don't be lazy.
It just means there's no sense trying to read the docs.
That’s entirely fair. I’ll certainly work on editing the docs and making them more legible over time.
Came for the "doomformer" based snow lyric puns... Left with a feeling my people are on slashdot.
Next step - take the cpu implemented in transformers implement it here and make it play doom. The compute needed still won't be able run Crysis well though.
oh my god i thought the crysis thing was a Turkish joke i didnt know it was international
As a fellow doom-weird-places-fitter, this is art. Congratulations :)
This is absolutely incredible. I feel like this could help visualize a neural network, but you have to know a lot about DOOM to really understand it.
you have to know a lot about lots and lots of DOOMs, in parallel
Self-plug I have a very similar blog post title but a very different approach. Honestly, I like your idea more than my own idea. https://nostalgia.dev/posts/running-doom-inside-a-transforme...
what if you run doom on doomformer?
I havent fully recursed but what I have done with this is use a DOOMformer to generate a level, have python compile it, have DOOMformer generate the name of the level, and use a DOOMformer brain to complete said level. I have also trained said DOOMformer (you can see the levels "weights" updating as it trains) to aim better lol
I think it might be more interesting to target Boom/Boom-compatible, if possible, rather than ZDoom. ZDoom already has so many advanced features, you could implement the entire thing trivially in ACS. At least a table of required engine features would be nice to see; some are likely already supported in other source ports.
the backbone implementation using ZDoom certainly wasnt difficult, but some aspects of scaling up were tricky. I definitely want to make it at least Boom compatible eventually
This is so trippy, very cool, love it
I made a classifier play doom[0]. But you turned the whole network into Doom! Amazing
More AI slop. No thank you.
amazing
Wow!
now everything can run on DOOM ?
you can theoretically build fable on doom