Thibaut Patel ☀️

@jessethanley @aarondfrancis @simonhamp I'd add: outside of any kind of batch computations (training on GPU, processing large datasets...), which were always a good thing to run overnight. I'm also skeptical of the current state of agents when they are autonomous for more than 5 minutes 🤔

@aarondfrancis @simonhamp Has anyone posted proof of stuff they are running overnight that is productive (yields value to their customers)?

Goal: be so productive when in front of the computer that you don't need to have a bot do stuff for you when you're not in front of it

@ben_cotte ❤️

seeing Mathias open up about mental health - and how skateboarding + a supportive community can help - really hits. Skateboarding becomes a pretext to share something precious. a common love for finding the perfect motion. an obsession with repeating the same thing, over and over. A shared nostalgia, too. and maybe most importantly, a place to feel heard and respected. I’m genuinely happy about what we’ve built. 🫶

moments like this are why Old Skaters exists. meeting people like Mathias makes all the effort worth it. watching how they push through life - how we support one another and how a side passion makes that journey stronger.

Free Chrome extension if you want to try it: chromewebstore.google.com/detail/retailr…

Killed a 2.5 year side project today. Selling spreadsheets on Gumroad wasn't it. Building RetailRank instead.

@JohnONolan @AlcoveNews ahah well never mind, off to restart my mac just in case!

@thibpat @AlcoveNews I cant repro on any browser or device, inc those 😅

RSS reader in full swing - this is such a bad idea Subscribe for early access: alcove.news Or follow: @AlcoveNews

@JohnONolan @AlcoveNews Firefox on macos, I have it too on Chrome on macos, it's happening especially when I type in the field, there lag between keystrokes

@thibpat @AlcoveNews huh - shouldn't be - is just a Ghost embed 😅 what device/browser? will check!

@EliotAndres Apache License 2.0!

Weights are here but I don't have a Modelscope account: modelscope.cn/models/Tongyi-…

Z-image, the latest open-source model in town, is about to be released. It's only 6B parameters but it beats Qwen-Image and Flux 1 👀

@DanielLockyer Looks like there's a bunch of packages affected helixguard.ai/blog/malicious…

uh oh

@gdibner Love this idea, I'm always surprised when folks say AI and mean LLMs.

Totally agree there is an LLM bubble. AI bubble probably also happening but less severe and more of a short term thing. "Hugging Face co-founder and CEO Clem Delangue says we’re not in an AI bubble, but an “LLM bubble” — and it may be poised to pop. At an Axios event on Tuesday, the entrepreneur behind the popular AI platform and community site agreed that bubble talk is today’s “trillion-dollar question,” but said he doesn’t believe AI’s future is at risk if the bubble bursts. Instead, as Delangue sees it, it’s large language models (LLMs) — like those powering ChatGPT, Gemini, and other chatbots — that are receiving outsized attention, and that attention may not last. “I think we’re in an LLM bubble, and I think the LLM bubble might be bursting next year,” explained Delangue. “But ‘LLM’ is just a subset of AI when it comes to applying AI to biology, chemistry, image, audio, [and] video. I think we’re at the beginning of it, and we’ll see much more in the next few years,” he noted." Hugging Face CEO says we're in an 'LLM bubble,' not an AI bubble | TechCrunch

What do you mean "Completion time: 6min", I thought I had to wait the full 24h, and it's already here!!

My first OpenAI batch mode! After trying out the eval product, I'm trying the batch. I love the 50% discount, and it's very easy to get started!

You can simply choose to view the world in a way where every obstacle is a blessing and every setback an opportunity. Don't wish for things to be easier! Do work to become stronger.

@bemmu I had fun, thanks for sharing this Veo3 experiment!

Updated the door game (renamed to "DOORED") with faster loading times, bugfixes, quicker transitions. #Veo3

@benjaminulmet @badimon Beautiful one, I love the pace of it!

I've been working on so many projects at Photoroom lately and some of them deserve to be shared! Brand new video made me, with @badimon & tomdimajo.com

@UriBram Stockholm syndrome 🤭

something I ponder is my relationship to Spotify. The app is so bad, sometimes the basic pause/play buttons don't do what they're meant to, I get programmatic ads that play twice in a row, etc. There is 0 barrier to exit and tons of other apps I could use. But I stay anyway.

I'm collecting the @github edge cases these days 🫠

@ben_cotte @matthieurouif @stokeCIub I will let you know if the next steps for sure 💯

@thibpat @matthieurouif My pleasure! I’m stoked for you! And tell me if GB API suns down, we still have @stokeCIub session OGs generated by it 🫡✌️

Big day for Photoroom! We’re announcing our first acquisition: Generatebanners, a powerful technology to scale image production with text. It’s our first M&A, but surely not the last one. We’re also launching a new product: Visual Ads Automation—a tool to generate creatives from a single template into millions of combinations leveraging generatebanners tech. In the future, creatives will be tailored to each user, scene, price, and locale. Our product makes that possible at scale.

@sotak rooting for this duo 🙌

How is it going - first day 2 new features I didn't want to work on due to complexity 2 bug fixes I didn't want to get into due to complexity

@sotak oh yes running the tests is really key for the tasks I'm using to tests agents right now, thanks for sharing your experience, it's super helpful 🙏

Yes, I am spending most of the time creating the Issues now. :) The more detailed the better. I sometimes put there what I think could work or where to look, just like I would do with someone new to the project. Then you just assign Copilot to it and it will start working. You can see the log how it thinks in a PR. It didn't work for me at first because I did not have actions enabled for the repo. (Support helped). Then if something is not right, just comment in the PR. I am yet to explore how it can run the tests on its own, etc.

So it begins