Hacker Newsnew | past | comments | ask | show | jobs | submit | udave's commentslogin

Im not sure if transfer learning is still applicable to these models, but if it is then we can just train the last few layers and get the model to ace in a related domain ( going from SWE -> Cyber ).


Hope it survive the next few years of codification of all of humanities unfinished cheap ideas.


Why? Letting it burn to the ground will bring about federated git forges faster.


This is horribly accurate.


As if Microslop didn't have enough unfinished cheap ideas already.


I love that phrase.


Once I tried looking up whether you can configure GitHub’s merge queue to skip MQ checks for PRs that have already completed CI against the latest main commit. I knew from how MQ works internally that this wasn’t possible, but I wanted to confirm. The AI overview on the page said “yes, GitHub has this feature,” and that answer raised my eyebrows. My mind instantly spun up 10 threads thinking through how it could be possible. I typed out how in the AI chat, and the next message said “No, it’s not available". I don't trust AI overview since then, but since its positioning makes it hard to overlook.


lovely website animation :D


Thanks! :)


lovely collection. Thank you!


Ive made my own agentic IDE centered around worktrees and containerization. it allows me to run multiple development environments on my machine with each development server running in parallel, allowing me to spin up feature branches and test them instantly.


seems like the same trick as behind labelling price as $99.9


Yeah, your work was so great that we are gonna fire just 3999 of you.


try distributing this spec amongst your team members, ask each of them to drive it to completion. no follow up edits. deploy to individual environments and then run a rigorous test suite against all of the deployments. see if all of them behave the same way.


They won't. So what? This is not how specs are used, no one is saying that they are a replacement for source code.


Let's be real, if you and I both ask claude to generate a feature on the same project, what are the chances that it spits out 100% replicated code? But if we are to build the project using a Dockerfile, we will get the same binary and the same image. Products around LLMs are non deterministic unlike compilers.


I can assure you that a fully deterministic and equally effective claude is possible to build. And yes, that would mean identical prompts would yield 100% identical output 100% of the time. It would still make the occasional logical or factual error, but it would do so deterministically. Would this solve any of the problems with building reliable programs using LLMs?


it's nondeterministic because we chosen it by having higher 'temperature' in settings. I bet if you run open weights model with temperature 0 and on the same device the same prompt and turn off parallelism you will have more deterministic result (excluding some floating point operations).


its a good idea in theory. But capitalists will make sure this does not happen because greed is never ending. today its AI, 10 years down the line it something else. Hence i think the right to be lazy is for a lucky few.


> Hence i think the right to be lazy is for a lucky few.

The freedom or ability to be lazy may be for a lucky few, but the right should be for all. A right doesn’t stop being a right when it’s not being observed, and when that happens it should be a warning to us all.

For example, access to clean water should be a right. But some communities don’t have it, and we should collectively help them. Due process is also a right, and when people are systematically captured from their communities and unilaterally stuck in a cell, we should collectively remove the power to do so from those who are doing it.


well not only that but the space to think about systems of power and possible realities (like the COVID lull) is antithetical to capitalist aims. leisure is doled out just enough to prop up consumer spending. if you restrain your desires you can buy back your time (FIRE) but that affordance is reliant on the status quo (others consuming so that your investments do well).


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: