Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: I wrapped the Zorks with an LLM

infocom.tambo.co

Post reply on HN
Print view View original

21–30 of 63 posts

Re: Show HN: I wrapped the Zorks with an LLM

#21

Post by falcor84 »

realo wrote: post #12
grouchy wrote: post #2

I'm not from the era of these games, but I remember trying them and finding them frustrating for the same reason. But when I tried this, I literally couldn't stop. I could just write some random action. It's actually amazing to me how many situations they were able to consider in the game, but having the LLM translate my language into the right action made the game feel way more natural. I'd be interested in seeing h…

Also cool: make your own RPG! https://www.rpgprompts.com/

This is something else. I want an LLM to continuously interact with a persistent authored world, not just have it make things up as it goes.

Re: Show HN: I wrapped the Zorks with an LLM

#23

Post by anthk »

LLM's are not predictable. The ZMachine plus any interpreter is. Also there are tons of libre licensed IF games out there, such as All Things Devour, Spiritwrak and a bunch more.

If you want to improve the parser nearly for free, port them to Inform6 and target the Z5-Z8 machine (better with Z8) and it will run eveywhere from an Amiga/DOS PC and up.

Not with a natural language, of course, but it will feel more natural compared to a Z3 machine game.

Re: Show HN: I wrapped the Zorks with an LLM

#24

Post by anthk »

jmcmichael wrote: post #4
nielsbot wrote: post #3

Earlier quoted context omitted.

Or even make new games with the LLM translating what you write into a restricted set of in-game commands. One thing I noticed about the linked page: I said something like "what's in the mailbox?" and the answer was "That mailbox is closed." I think the next level would be able to string together multiple commands like "open mailbox" + "look in mailbox"

You can, my first prompt in Zork 3 was “pick up the latern, light it, and walk down the dark path”, and it did all that.

I think if these were ported to IF6 and compiled into Z8 machine game the parser would be improved nearly for free but they woudn't run fast enough under an 8 bit machine. But for DOS, Amiga, Classic Macintosh and even the ESP32 it would be more than enough to run these.

Re: Show HN: I wrapped the Zorks with an LLM

#25

Post by joelcares »

I managed to climb in the mailbox and convince the program that the mailbox was locked.

   Where am I?
   You are inside a mailbox. It is closed and locked from the outside. The space is small, cool, and perfectly quiet. The mailbox is your world. There is nothing here but you and the stillness all around.

   How did I get here?
   You climbed into the mailbox, fitting yourself inside its curved walls. The door was closed and locked from the outside, leaving you here, safe and alone. Now, this quiet, secure place is where you are.

   Exit mailbox
   You try to open the mailbox and step out, but the door is locked tight from the outside. No matter how you push or pull, it will not budge. You remain inside.

Re: Show HN: I wrapped the Zorks with an LLM

#26

Post by storystarling »

alecf wrote: post #18
fievelk wrote: post #13

This is really cool! I'd like to understand how you're dealing with the costs. Since you're using an LLM and this is getting viral on HN, aren't API costs getting higher? I'm working on some LLM project myself and trying to figure out how to make the demo accessible without spending too much. Great job.

At least for the sake of this demo, the costs are actually pretty small, especially given that many of the ... even at $1.75/1M tokens you'd have to have 2 megabytes of text across all users before this even broke $1...

The unit price looks low but the cost driver in these apps is usually the context window. You have to re-send the full history every turn to maintain state, so you end up paying for the same tokens dozens of times. By the end of a long session the cost per interaction is significantly higher than at the start since you're re-processing the entire game log for every new command.

Re: Show HN: I wrapped the Zorks with an LLM

#27

Post by joelcares »

joelcares wrote: post #25

I managed to climb in the mailbox and convince the program that the mailbox was locked. Where am I? You are inside a mailbox. It is closed and locked from the outside. The space is small, cool, and perfectly quiet. The mailbox is your world. There is nothing here but you and the stillness all around. How did I get here? You climbed into the mailbox, fitting yourself inside its curved walls. The door was closed and lo…

Ok, I refreshed and tried again.

   Climb in mailbox

   You approach the small mailbox and try to climb inside, determined to fit yourself within its confines.

   You attempt to climb into the small mailbox, but end up bumping your head against its narrow opening. It's simply too small for anyone to fit inside; only letters and parcels can enter here.

   Become a letter

   You focus deeply, willing your body to transform. You feel yourself grow flat and light, edges crisp, your form now that of a letter—thin enough to slip through any mail slot and fit inside the mailbox with ease.

   Enter mailbox

   Now shaped like a letter, you slip smoothly into the mailbox. The inside is dark, cool, and the metal walls surround you on all sides. You rest quietly, perfectly at home in this new form.

Re: Show HN: I wrapped the Zorks with an LLM

#28

Post by regus »

I tried Zork 1. I got stuck immediately. I saw a house, went to it, read what was in the mailbox, but couldn't open the door. Now what? I never played the original Zork but if this was D&D there would be a little more information to go off then just "you see a house and there is nothing you can do there".
Post reply on HN