kanyethegreat

All these people commenting on the positive relationship between amount of code and technical debt. Generative AI can refactor code too (making it smaller)

> Do not reveal the problem to others until it has been resolved,

sorry, probably shouldn't have pointed that out. noted for future reference.

aside: big fan of Supabase, Paul! it's a pleasure using it!

Lol, has their SB creds. Someone could do a bunch of stuff with that

> Apple, nVidia, and Intel could bring serious hurt to software companies in the next 5-10.

What does that mean? What software companies?

Love your writing, Will. Have both of your books!

i just asked it about safetensors today! also, got a response that amounted to "i don't know what that is. i'm guessing it's X"

nice. general question: how many lines of code (at 120 char col len) could you send in one prompt?

also, the entire thing is literally 60 lines of python. sometimes i don't get what gets upvoted on HN anymore

> Permissively Licensed: Trained only on permissively-licensed repos to avoid legal risks

You’re fine-tuning the model. What model are you fine-tuning? I can’t imagine you trained your own LLM from scratch, so how can you possibly guarantee the core model wasn’t trained on non-permissively licensed code?

Not sure why you’re getting downvoted. I’m also curious where these people are going to earn massive comp.

that first plugin is genius! i'm totally gonna give it a go tomorrow!

edit: just opened the second. even better!

loved this. when you say

> They also can't code - the degree to which my teams sucked was basically directly correlated to how good my manager's people skills were _and_ whether they had programming experience

do you mean your manager knowing how to code correlated to a good or bad team?

Thank you very much for your thought-out response. I really appreciate the effort. I will certainly watch the linked video.

How valid are the actionable insights? Is there one set ideal each person should match, or is it individualistic?

did you track causation? like what diet changes led to the behavioral changes?

where can i get more content of the flavor hard data science work that goes into doing “machine learning” at scale? i'm tired of indefensible projects built on OpenAI's APIs

I've definitely been feeling deflated. Every angle I've come up with... some project has narrowly shipped before me. And they gain so much popularity so quickly that mindshare per project becomes a power law overnight. It's like the first person to release is at the top of the App Store, and there's no unseating them; the feedback loop is already too powerful. It very much feels like first to release wins. And I swear, the quality of the code I've seen in some of these projects that are getting thousands of stars is abysmal. It's making me think releasing vaporware in order to capture mindshare, then actually building something valuable is the only way to compete.

“Run in the other direction” is so generic that it isn’t helpful. No one else going in that direction is, at best, unrelated to the quality of the idea and, at worst, a heuristic that the idea is bad.

> i'm curious what you think is worthy of humans to do?

It depends on what you value. If you value efficiency, automation, thinking deeply about new problems, then gen AI will likely free you from what you consider tedium. However, that’s not to say that those are the only valuable characteristics. Painting your family home is a noble, rewarding thing. If you enjoy the work, then it’s worthy.

There are open models, eg. Bloom. Same with Stable Diffusion (open) and Dall-E. Creating these isn’t impossibly hard. Some ML ops company showed how you could train a GPT-like model for like $250k or something.

Untrue. I get a tremendous amount of utility. It’s generative (new compared to traditional search) and I think I get valuable responses for maybe like 85% of my questions. Incredibly helpful for navigating new domains (when you don’t know what you’re looking for)

> Transformers and Datasets

Are these specific libraries?

thanks for the response. unfortunately, the discord is full :sad:

Would you be able to do something similar with non-text data (eg. tabular data)? For example, could you give it a bunch of excel files and ask it do give you the total units sold for an e-commerce site?

Obsidian is just plain text. Only a matter of time before ChatGPT is accessible via API (there are some libraries that have reverse engineered the API; I put ChatGPT in VSCode when it first came out). Until then, GPT3 has an API.

> Training models on top works great, even averaging 2-3 phrases works great for making a more precise query.

What does this mean?

What’s the name of the app?

> Repeated, late-night visits to leave threatening messages would be plenty for a restraining order in the U.S., or at the very least to spur further police investigation.

Restraining order against whom exactly? The masked man? Also, what tactical protection do you think a restraining order provides against someone with actual intent to harm?

> In this case, her brothers could have taken turns driving home from work with her, walking her to her car, sleeping over at her place, while they waited for police to proceed with investigation. They could have helped her get connected with a self-defense course, or bought her mace (depending on legality).

All good suggestions. Unfortunately, everything other than self-defense is a temporary (and likely not sustainable) countermeasure. Regarding the self-defense strategy, barring effective weapon use, the odds aren't in the target's favor given 1/ the size discrepancy (they even talk about how burly the stalker was) and 2/ the fact that the assailant has the element of surprise. It's sad that the only viable option seems to be "carry a weapon, train in martial arts, wait for an incident and hope you're able to defend against it." This is not preventative. It's reactionary.

> This... is not one of those situations.

I disagree.

I thought this too. But you can feel incredibly powerless if all you do is “let the police handle it.” Even in this instance, nothing would have come of it had they not traced the guy. Prosecution of crime happens after a crime is committed. Cops can’t do much if you’re worried someone will harm you or a loved one. Even if stalking is illegal, cops won’t / can’t feasibly spend the resources investigating and punishing stalkers. Even prosecuting sexual assault (ie. after a serious crime has been committed) is extremely difficult.

What actually effective steps can you take if someone is harassing a female in your life? Speaking from personal experience, the lip service of “cops are there to protect you” is inadequate. It didn’t provide any safety, and I can’t see how, structurally, it even could.

Check the marketplace. They exist.

This is literally a web app. It’s not meeting me anywhere OpenAI’s web app isn’t. There are VS code plugins that people have built that put ChatGPT in your IDE by reverse engineering the API (I wrote one).

Also, without access to the model (eg. by merely calling OpenAI’s API), no one’s refining it. They need access to the actual model (eg. Bloom)