It's pretty crazy how some video by a relatively small content creator snowballed into Cloudflare making meaningful changes and addressing platform issues
LLMs are only going to improve by fragmenting them into specialized systems for low parameter high performance results. We’ve reached the point where models will get smaller and more compact
I thought "The Bitter Lesson" was that whole a specialised system will outperform in the short term, generalized systems with lots of data win in the long term.
Over time. But for a given instant, specialization will always win. That message is for researchers, who seek to have long term impact and it's bitter because it goes against their desire to provide long term impact from their own clever abstraction or insights.
But it's informative for the engineers that need something right now, because it means taking the best general purpose tool and specializing it will outperform the general tool, and you can sustain that if you are willing to always hop tools and respecialize. As we may.
I think there is a bitter lesson to the bitter lesson.
Sure you can throw more compute at it. But it cost a lot of money and you hit resource limits.
We have been doing an end run around the bitter lesson with prompt engineering. Also by using different models for vision vs. text. By getting (human coding) agents to "think" and run code.
The bitter lesson might be that you cant predict what the thing is that will be most optimal tomorrow and any player in the AI game can be innovated out of existence at any time.
Like I wanted this from a year or two ago to just lets say have a model which lets say is genuinely really really good at sveltekit as an example instead of a model which is good at a lot of different things of sorts yknow
A model for sveltekit, A model for react and for coding general purpose too and preferably we can have a website which can make it easy to find these models/run them, ollama comes to my mind right now but it has really enshittened a little bit from the time when I was thinking about this but so maybe now a little competition on that side wouldn't hurt I suppose.
Thanks so much for the encouraging words; I hope any devs who try it out will fix the pain points I have when writing my own SVGs. The project resonates from a love for SVGs so I hope any developers with that same love would be happy to follow along!
Tried the demo on desktop. Processing took a few minutes for a standard JPEG, and when it "finished" the Download button redirected to an API endpoint that returned "invalid params," so I couldn’t save. Do you have client vs server details you can share? A visible progress indicator and an error toast with a retry link would help debug.
The UI is pretty cool; it could maybe benefit from a sort of breadcrumb tracker at the top for a more linear flow-state but the data visualization is seamless.
Notepad has always been about simplicity. Adding AI feels like bloat opposed to a feature users were asking for (or would ever want.) Local models are fine, but tying them to Copilot PCs or subscriptions just fragments the user base. If Microsoft wants this to stick, they should make it optional, and stop slowing down the tool people actually rely on.
Notepad had already become a moving target of inconsistency, no longer able to maintain place after scrolling down, nor when toggling Word Wrap or adjusting window size.
Still-changing deviations in expected tab and simple format structure between emerging versions . . .
Up until the last few years you could print any text file saved by any previous verion of Notepad for decades, and have it come out the same.
Nothing of the sort any more.
When the most useful feature of such a successful legacy app is that it never changes :(
I've said it before, technology has come a long way but people used to be more advanced. Especially at things like this.
Looks like they've built an entirely different tool now, and didn't have the guts to let it sink or swim on its own.
Amazing project; I love the ambition and the TUI is gorgeous. Very neat! Mind if I ask; does the project have some analytics functions for logging network data? Thanks!
