WildGreenLeave

The problem is that with max 1 token it is not 100% guaranteed that it will be a yes/no or a score. This model (and engine) solves that by taking the probability of the yes/no token. For 1 question it is faster but not a whole lot, the power comes from being able to ask more then 1 questions for merely a few ms extra per question.

E.g. 1 question would be 300ms but 30 questions would be 360ms total. Most of the time comes from the image decoding.

As a proof of concept I've been working on a Jev style vision model that can rapidly answer boolean and choice questions based on images (See [0] but not the point of this comment). One of the reasons I wanted to make this concept is because I was also thinking about exactly what the parent comment suggests. I just haven't gotten around to getting (3d printing) a robot arm yet :)

[0]: https://huggingface.co/MeerDevelopment/Qevi-2B

I have driven a lot in Europe and a lot in the night (I like to avoid traffic jams and use the no speed limit on the autobahn), I have to say that this does almost exclusively belong to Germany in my experience. Which makes perfect sense because Germany is in the center, has a lot of truck traffic and long autobahns so it makes sense the parking lots are limited. I have not seen this in any other country other then quite exclusive circumstances.

I tried OpenCode but didn't particular like it as a Claude Code user, that is the main reason I switched to Pi. The reason I am sticking is how simple it is to extend it. I moved from Claude Code to Pi and within 2 hours (and the help of Claude Code) I have a setup that matches Claude Code and is even better for my setup.

Things I've added:

1. Built my own AI judge for 'auto' mode that matches my setup.

2. /plan /go for planning and executing.

3. /flow for A-Z setups. That includes planning, executing, testing and shipping.

4. /deep-research a multi fan-out setup for researching a topic.

5. My own sub agents.

6. A TaskCreate/Update/List setup.

7. Monitors.

8. BashOutput / KillShell.

9. Proper notifications with Notify that uses macOS banner and work.

10. Spawn tool that triggers multiple sub agents.

11. A bridge between signal to use Pi remotely.

Yes a lot of these things is something that was already in Claude Code but now I don't have to use Claude Code and I can customize it to fit me exactly.

Separate companies but Dennis (founder of Ploi.io) is the other co-founder, so indirectly linked. :)

Appreciate the suggestion, thank you! Definitely not charging that much :-)

Wow, I have to admit that I have not heard anyone in the past 2 years or so to be on Heroku so it makes sense. I think they handled it quite well knowing that there most likely have been a steady decline of users.

Generally I would avoid promoting myself but in this situation I think it fits the topic. I'm co-founder of a Platform-as-a-Service based in Europe named Ploi Cloud [0]. We focus on web applications working on NodeJS and PHP but would be open to other platforms if people need it. Heroku has always been a source of inspiration to me so if you are looking for an alternative and care about it having a strong European presence please check it out. (We do have a US location too!)

0: https://ploi.cloud

I've had the idea to setup an AI that automatically (re)designs a room using IKEA stuff. It would definitely help me decorate my room in a better way.

I'm part of a new hosting platform focusing on the European market, would you be open to discuss what you are looking for and see if we can become partners? Obviously no strings attached :) If so, please drop me an email zander@ploi.cloud

Edit: Forgot to add my thought; I think you should be more clear about what Clink does compared to other software like Claude Code. The building demo's to me (a developer/devops person) are cool but my first thought was 'Why would I use this over Claude Code'? And only in this thread I saw the USP for Clink which makes perfect sense for non-techies!

It is missing the (to me) most important part. The reason why these changes are made.

I think it is a great tool, as someone who has this on his bucketlist but no experience it is really handy. Few points of feedback:

1. When I select the start date, maybe autofill the end date with 2 weeks or so. 2. I dropped my email, but that is not something I enjoyed doing. 3. I think there should be a clear reason what is expensive and what not. My 2 week itinerary was 25k. I have no idea if this is expensive (probably not), but to me this feels insane.

As a European missing a managed hosting solution, me and a buddy of mine are building an alternative: https://ploi.cloud

The goal is quite simple, allow developers to host their application with easy straight forward pricing. We are about to launch very soon. Everything is built on Laravel/PHP.

We are open to beta testers, so if you feel you want to test this please drop me and email in my profile.

I'm a heavy PHP & Laravel developer and I speak for myself and a few close friends around my in my network who are like me. We all consider Laravel the reason we are still within the PHP scene and didn't move away. So in a sense I think it is true.

That said, the recent changes around Laravel (being bought out and becoming more and more commercial) is not something I (we) consider a good thing. Not necessarily a bad thing, but we all know that a OSS framework becoming commercial doesn't usually end well.

The first thing I setup when I started to manage my own Kubernetes cluster more then a year ago was this Warrior, I completely forgot about it until this post.

Has been active for over a year steadily working the recommended project. Downloaded over 3TB in 6 days (node reboot, so pod was restarted and stats are not persistent). So rough extrapolation is about 180TB. Happy to help the good cause of the ArchiveTeam!

Edit: typo

Correct me if I'm wrong, but, if you run multiple inferences at the same time on the same GPU you will need load multiple models in the vram and the models will fight for resources right? So running 10 parallel inferences will slow everything down 5 times right? Or am I missing something?

I'm inclined to say screen.studio [0]

0: https://screen.studio/

Most people are within BMI range, I for one, actually enjoy the benefit. And as long as the majority of the people benefit from it I prefer it this way then the other way around.

As you have figured out this change was only about the liquids to adhere to EU standards, I of course dislike it as somewhat frequent flyer, but it makes sense to make a standard.

Now if only I could leave my laptop in everywhere, that would be nice. I never fly without my laptop and it gets old quite fast somehow having to figure out if I should take it out (and spent 5 minutes putting it back in, which sucks if it was not required) or try to take it out when people are waiting on you. I just wished airports would put clear signs up what you have to take out so I can prepare before I get to the belt.

I run everything on Longhorn, including databases. Not the highest iops but definitely worth the hit for the easy of migrating.

My personal tip is to slow your breathing, slow it down enough that you breath too slow and feel a bit deprived from oxygen. Then slowly breath faster but try to keep it as low as possible.

I think (but I am not an expert in this) it slows down the heart and with that, allows you to fall asleep easier.

You don't _have_ to, I think previous commenter means that after a few days of feeling tired and groggy you will be tired earlier in the evening and go to bed earlier.

Jetbrains is the perfect example of a subscription based platform that I will happily pay for without any bad feelings. I pay yearly, after a few years you do get a discount (so I am a happy camper). And if at any point I decide I don't want to pay it anymore, I keep access to the latest version of the software I paid for.

As far as I know this is a true American problem, I have never heard about this, or have experienced this in Europe or Asia.

On iOS long pressing a link so it previews is very nice for these kind of things. I only wished you were able to scroll inside that popup/preview.

I think this is partly because there is not really an alternative once you run out. Getting a jerrycan of gas from a gasstation is possible, a reasonable expectation can be made that if that happens a good samaritan will help you. Or a friend can drop by with one.

The same thing cannot be said about electric cars, unless you have a good friend with a fuel generator. Even then you still have to wait for it to charge. More likely you will be towed, at least in The Netherlands they will tow you to the next parking place. If you are lucky it will be one with a charger, if you are unlucky you are once again on your own.

I'm not a professional myself either, but the things I would check are:

- Make sure you run hardware acceleration, you can see if this is applied after opening a connection and checking the details. Without acceleration it will be terrible. This can either be as simple as enabling it or possibly you need the monitor below.

- Make sure a monitor is plugged in (there might be a way around this, but I just bought a $4 HDMI dummy instead of trying to figure out how to do this the software way)

Edit: formatting

I've been using Parsec on my local network for several months now and I see absolutely no latency whatsoever. GFN and any streaming provider gives me a very small but noticeable lag in shooters, but Parsec on my home network is genuinely perfect. I game from my Macbook and I play whatever I want from my homeserver (3090 and 4060) in another room. Both my Macbook and computer are wired of course. Important to note is that I needed a dummy HDMI on my computer for a 'true' screen, any kind of virtual screened caused it to lag hard, probably because the game or stream ran without any kind of hardware acceleration.

I have to say, I love the way Parsec works and allows you to use your whole library (since it doesn't stream the game, but it streams your whole desktop). But if there is ever a service that allows you to get the same thing including remote servers (since I travel), I would probably switch.

Maybe a geolocation issue, but this one keeps on redirecting me to a whole different URL [0] with no taco face :(

0: https://www.yum.com/wps/portal/yumbrands/Yumbrands/company/o...

I often experience ChatGPT given outdated information which is wrong at this time, but wasn't necessarily wrong in the past. One big example is a popular framework for Laravel called Filament [0] which released a V3 a few weeks ago. I don't even bother anymore with it because it is useless. However, for example 90% of my DevOps tasks (mostly Kubernetes) are partially done with Kubernetes, either explaining impact or even writing manifests. It is genuinely awesome for that.

[0]: https://filamentphp.com/

A starting point for that might be the way I am doing it. I am using LocalAI to expose OpenAI compatible endpoints and then get access to those via Tailscale.