Michael Grenier

@jeeacheff @claudeai Gotta be provider-flexible nowadays

@claudeai The upcoming model better be a banger, ~98.8-99.2% uptime is quite subpar right now.

The way I build the vectors is slightly different: The screen is divded into cells. Then I divide those cells into a subgrid. The average luminance of each cell in the subgrid is what I use to build the vector I think the math also differs in some places because it all runs in compute shaders alexharri.com/blog/ascii-ren…

@rqobela C'++, VBA, Rust, Go, PHP and many more. Like to learn stuff but my org is all in on dotnet

Programming language you learnd but never used again is...?

@anthonyriera Neck pain

what stopping you from coding like this?

@Saadeghi A pov: PDF = money + data B pov: PDF = free Ad = money + data As a pro-privacy consumer, blocking adj prevents the website from leaking my data to a third-party. The fact that also stops the website from getting money is collateral. Lock paid content behind a paywall!

Website A sells a PDF book. You pay, creator earns revenue. Website B has the PDF book for free. You see an ad, and the creator earns revenue. How is blocking ads on website B any different from pirating a file from website A? Both actions prevent creators from earning revenue.

@Saadeghi @mbvisti Call it fees or taxes, it doesn't cover the cost of asphalt. It needs a lot of taxpayers from multiple sources. The website isn't getting revenue from others website ad revenue.

@grenierdev @mbvisti Those private companies pay a big fee to the municipality for every single billboard that is visible to the public.

@Saadeghi @mbvisti Wrong analogy. It's the taxpayers that paid for the asphalt. Billboards are owned by private companies.

@mbvisti Yes you can. That's how you can use everything on the internet for free. Because everyone is earning money from ads. Not just the internet. When you walk outside of your house, the asphalt you're stepping on was made possible because your city earns from billboards and ads.

this takes 30 seconds to fix

We published a CVE for a moderate dev server security vulnerability. Please update to the latest versions: 6.2.4, 6.1.3, 6.0.13, 5.4.16, 4.5.11. github.com/vitejs/vite/se…

🔥 Almost done with the registry schema. It’s unbelievably powerful for code generation and distribution. Define code as a flat JSON file and distribute it via the CLI. ✅ Custom styles: bring your own design system, components & tokens ✅ Extend, override, mix & match components from third-party registries and LLMs ✅ Install themes, CSS vars, hooks, animations, and Tailwind layers & utilities On install, the code adapts to your local styles. ⬇️

@threepointone I was curious to see what the PR looked like. Damn that's bad! github.com/threepointone/…

If you’re going to submit an ai PR to my repos but you haven’t actually checked whether it works

@thekitze If you're publishing a package of react hooks, with minimal JSX, you can easily convert them to JS. Example #L146" target="_blank" rel="nofollow noopener">github.com/baseless-dev/b…

how did I learn about jsr just now 🤦 it makes it so stupidly easy to host and update a package of ts code it's crazy. now I can finally publish more stuff!! (wish it supported jsx tho)

@mattpocockuk Something like CloudFlare AI Gateway : prepass with all the functions (name+desc only) and user prompt and ask the LLM to filter out which might be useful. Then send the relevant functions and user prompt.

What tricks have you used for getting round the soft limit on the number of tools that a LLM can use? - Making tools more polymorphic - Orchestrator LLM's What else?

@mayakyler I get stuck on "not shipping" by cramming every feature I can think of...

Unpopular opinion: Coding skills are a disadvantage for SaaS founders. They get stuck on tech and often ignore marketing.

@stevemeans_rlsy Just people I followed

@grenierdev Did you remove also some followers?

Have you ever unfollow everyone to start from scratch here on X?

I have an immortal notification. Am I the only one that has this bug?

@OpenAIDevs Poor choice of color palette. Makes me question if I'm color-blind or not!

It's faster than `o1-preview`, using 60% fewer reasoning tokens on average. We've also added a new `reasoning_effort` parameter, which lets you control how long it thinks before answering. The snapshot we're shipping today is `o1-2024-12-17`—see how it performs:

We're bringing OpenAI o1 to the API. We're rolling out access to developers on usage tier 5 starting today, and rollout will continue over the next few weeks. o1 supports: ⚙️ Function calling 🗂️ Structured Outputs 👀 Vision 📝 Developer messages 🧠 Reasoning effort

@pilcrowonpaper 🤯

@grenierdev You can keep guessing the same code over and over again. The odds are still 1/1,000,000 every time

Even if you rate limit TOTP verification to 10 attempts/hour, you have a 10% chance to crack it in 1 year. If someone has thousands of IP addresses and your app isn't accounting for that, it'd be cracked in a few days