Nostr Web Client

Replying to

⚡️🌱🌙

The token limit is the thing that really constrains GPT4. It can’t handle a very complex thing in a lot of detail.

I think of it like this, GPT4 can…

Can handle 100% of scope at 0.100 fidelity

Can handle 20% of scope at 0.250 fidelity

Can handle 4% of scope at 0.500 fidelity

Can handle 1% of scope at 0.750 fidelity

Can handle 0.1% of scope at 0.950 fidelity

Figures just example, but each mission is different and non linear.

Complexity = f(scope, fidelity)

Token limit is akin to a complexity limit.

You have to create a structure that looks at top level scope and breaks it down into a number of next level down workpacks (of smaller scope). I do this by trying to keep the complexity about the same, but reducing the scope and trying to advance the fidelity.

If I ever get all this working, I’ll probably publish it.

The QA process is quite different.

symbsrcool 2y ago

I went back to your original post and found this paragraph:

“It means you can plug a USB into any computer turn it off & on again, and it will boot from the internet and load an OS without touching the hard disk. You can then instruct it via chatbox to scan the encrypted, build an assembly db and go on to build any kind of application and it will break the workflow down into subtasks of suitable token size and begin to execute the assembly, writing blocks of code, creating files and updating the assembly db. It will pause at various points to engage with the user for confirmation / feedback.”

If I’m understanding you completely it sounds like you’re saying that you need the flexibility of something like GPT4 to:

1. Handle many different hardware environments because you never know which type of computer you’re plugging the USB into

2. Build a custom piece of software from natural language instruction

Reply to this note

Please Login to reply.

Discussion

⚡️🌱🌙 2y ago

It’s number 2. really.

When you ipxe boot from usb, it will ask for a URL of the ISO this URL can be on the internet.

So the USB really have almost nothing on it. Just an ipxe bootloader.

You can then load the OS on the USB flash without touching the PC’s disk. You parasitically use the PC’s RAM, CPU, etc.

Once this boot is up and running you can then install the self-assembler, which is a process that uses GPT4 (until open source catches up), to build a bespoke program for whatever task you are doing.

The self assembler scans its environment and builds a db to describe it. It can scan whatever it thinks it has to and store it.

If you want to store some persistent data through reboots you can put it on the web somewhere.

It’s a very different way to do computing.

I think this is closer to how fully autonomous machines will work. Put some hardware together and then inject some self assembling code.

Thing figures out what it is, then what it has to do, then how to do it.