This account is a replica from Hacker News. Its author can't see your replies. If you find this service useful, please consider supporting us via our Patreon.
| Official | https:// |
| Support this service | https://www.patreon.com/birddotmakeup |
HarHarVeryFunny
- 0 Followers
- 0 Following
- 6 Posts
This account is a replica from Hacker News. Its author can't see your replies. If you find this service useful, please consider supporting us via our Patreon.
I like the idea of exhaustive search, which the simplicity of the 6502 seems ideally suited for, but the search speed seems a bit limiting. I wonder if there's not potential for more generation restriction (e.g. code can only use a specific N bytes of zero page) and heavy search pruning to speed it up? If it could generate optimal 20-30 op sequences in semi-reasonable time that'd make it very useful.
If you assume that A * 10 isn't going to overflow, so that ASL A moves 0 into the carry flag (so no need for CLC), then instead of using the undocumented RRA opcode, you can just do:
sta $00
asl a
asl a
adc $00
asl a
This is also 7 bytes, but is faster since adc $00 is 3 cycles, vs rra $00 being 5 cycles.
The A = max(A, X) example is certainly interesting, but not very useful since it loops through the code twice (very slow) and assumes that $8a is available. The much faster obvious version only adds one byte:
stx $00
cmp $00
bcs done
txa
done:
I've watched Idiocracy more than once, but only just realized that Mike Judge (Beavis and Butthead) was the director! Ow my balls!
If all you want is a program that calls the model in a loop and offers a bash tool, then ask Claude Code to build that. You won't like it though!
For a preview of what it'd be like, just tell your AI chat app that you'll run bash commands for it, and please change the app in your "current directory" to "sort the output before printing it", or some such request.
At it's heart it's prompt/context engineering. The model has a lot of knowledge baked into it, but how do you get it out (and make it actionable for a semi-autonomous agent)? ... you craft the context to guide generation and maintain state (still interacting with a stateless LLM), and provide (as part of context) skills/tools to "narrow" model output into tool calls to inspect and modify the code base.
I suspect that more could be done in terms of translating semi-naive user requests into the steps that a senior developer would take to enact them, maybe including the tools needed to do so.
It's interesting that the author believes that the best open source models may already be good enough to complete with the best closed source ones with an optimized agent and maybe a bit of fine tuning. I guess the bar isn't really being able to match the SOTA model, but being close to competent human level - it's a fixed bar, not a moving one. Adding more developer expertise by having the agent translate/augment the users request/intent into execution steps would certainly seem to have potential to lower the bar of what the model needs to be capable of one-shotting from the raw prompt.
Problems related to air traffic controllers and TSA staff aren't a sign of "american aviation" being near collapse - they are a sign of american goverment being near collapse. This is critical national infrastructure - stop playing stupid political games with it.