- Rabbit R1, AI gadget, runs on Android app, not requiring “very bespoke AOSP” firmware as claimed by Rabbit.
- Rabbit R1 launcher app can run on existing Android phones, not needing system-level permissions for core functionality.
- Rabbit R1 firmware analysis shows minimal modifications to standard AOSP, contradicting claims of custom hardware necessity by Rabbit.
Not sure what the surprise is. It is a device that needs you to sign into an account and have an internet connection. Ie it is just a dumb terminal. Kind of like how Alexa speakers are useless without an internet connection. All the processing is done on servers.
Even if it was not an android app, you probably could have made a clone of it since you are basically interacting with a web api.
In their defence the device has the advantage of giving you instant access to their ai through the touch of a button whilst on phones you would need to physically open the app as the “assistant” functionality is already reserved to Siri, Bixby, etc.EDIT: I wasn’t aware you can change assistants.Personally I like the concept if the processing was done within the device, but considering you need monster machines to run llms I guess we are least decades away from that reality.
Well you can change your ‘assistant’ in Android so not even that.
Ah wasn’t aware of the ability to change assistants. I stand corrected!
Technically, every Android phone uses a “very bespoke AOSP”, because the Android kernel is customized for the hardware of every single phone model, which can include things like hardware drivers and carrier services.
This is the reason that there is no universal Android ROM that works across every Android phone, unlike Windows or normal GNU/Linux distributions.
What about custom ROMs like Lineage? The only thing holding it back from working on every phone is that many phones have blocks to prevent installing a custom ROM in the first place. Could be just like windows in that it has every driver for every piece of hardware in the package, just bloating it unnecessarily.
The manufacturer has to release the phone’s kernel source code before any custom ROM development can happen for the phone most of the time for that reason.
There is a reason that GrapheneOS only works on a couple of Pixel phones.
Could be just like windows in that it has every driver for every piece of hardware in the package, just bloating it unnecessarily.
Google specifically designed the Android kernel so that the driver are excluded, unlike the normal Linux or Windows kernel, because, long story short, Qualcomm did not want it to happen.
Every device had it’s own device tree and kernel with custom driver’s, binary blinds, and system software. All of it runs beside the closed source modern os
I’m not entirely sure because I’m not very knowledgeable about CPUs, but it seems this is largely a problem with ARM architectures and their lack of standardization, isn’t it?
There’s nothing like uefi between the os and the hardware
Another day, another scam, nothing new.
I’ve no skin in the game related to this device or software. I simply don’t care. However, in terms of an AI assistant, I am curious if there is anything on the market, including this APK device, that is worth using. For someone who is privacy centric, advert avoidant, and security focused… does (or even can) such an app/tool exist?
You won’t like anything on offer currently except for that which is entirely self hosted.
What are a couple of the best self hosted options? I wouldn’t mind giving it a go on my server. I may not have enough juice with what I currently run, but perhaps as a proof of concept first and then new hardware later.
What db2 already said. Microsoft just released Phi-3 mini, which could, allegedly, run locally on newer smartphones.
If I understood correctly, the Rabbit thingy just captures your information locally and then forwards it to their server. So, if you want more power, you could probably do the same by submitting the same info to a bigger open source model than Phi-3, like Llama 3, hosted on your homelab. I believe you can set it up with huggingface/gradio, which sort of provides an API that you could use.
That way, you don’t need a shitty orange box, and can always get the latest open source models with a few lines of code. There are plenty of open source frameworks in the works at the moment, and I believe that we’re not far off from having multi-modal LLMs running on homelab-level hardware (if you don’t mind a bit of lag).
huggingface! i found it once and never could remember again who hosted all those models. thank you!
i use a different device than homelab, but now I am curious what I may be able to achieve with my syno system. it’s hw weak, so probably not a lot. but i would like to give it a go. if it’s decent, i may consider another device for the purpose.
Good luck! You can try the huggingface-chat repo, or ollama with this web-ui. Both should be decent, as they have instructions to set up a docker container.
I believe the Llama 3 models are out there in a torrent somewhere, but I didn’t dig to find it. For the 70B model, you’ll probably need around 64GB of RAM available, but the 7B one should run fine with just 8GB. It will be somewhat slow though, compared to the ChatGPT experience. The self-attention mechanism can be parallelized, which is why you will see much better results on a GPU. According to some others that tested it, if you offload some stuff to RAM, you could see ~10-12 tokens per second on an RTX 3090 for certain 70B models. But more capable ones will be at less than 1 token per second, all depending on the context window you use.
If you don’t have a GPU available, just give the Phi-3 model a try :D If you quantize it to 4 bits, it can apparently get 12 tokens per second on an iPhone haha. It should play nice with pooling information from a search engine, or a vector database like milvus, qdrant or chroma.
I don’t really care for this product. It’s another unnecessary AI assistant. What I’m struggling to understand is why it matters which platform it’s been built on. What difference would it make if they wrote an entire OS from scratch only for this device instead of using Android, if the end product would be the same?
Because the main criticism of this class of products is “why in the fuck would I need a device for this? my phone already has a data plan, a microphone, and a camera. Make it an app” and the response is some vague “oh well it’s so advanced (it’s not.) it couldn’t possibly run on a phone”.
The vision is that once TPUs become affordable enough to run these models on-device, you would need a device that has such a TPU and you would go to them. But this is completely overlooking the fact that all snapdragons and the like would also have the same TPUs integrated, and also we’re not there yet, so for as long as you need to send the query to openAI’s API, why is this not an app?
It just drives the point home that it should have, and could have been an app on your phone. If they hadn’t sold this as some sort of revolutionary new product and instead were honest to themselves and us about what it was this would be a different story.
Just like all tech bs it’s just a bunch of lies to make initial sales that they never follow through with.
But maybe people would like a standalone device, without the distraction a regular smartphone brings.
I’m not defending this thing as AI in its current state is near useless, bar some niche applications. But the decision to make it a standalone device isn’t really controversial in my opinion.
I think that you’re in the minority with that opinion then. I don’t want another thing to carry around. And I don’t see how this would be less of a distraction seeing as it doesn’t do anything useful. To me that makes it more of a distraction. If you want to not look at your screen to perform tasks that’s already an option with your phone using Siri or Google’s assistant.
No, you don’t need a ‘very bespoke AOSP’ to turn your phone into a Rabbit R1
I just fucking need to get the APK from a goddman reliable source and I only know APKMirror
The article says that you need a custom launcher to establish a connection to their cloud service, so the apk only won’t do it either way. But give it a few days and someone will have rigged up a full package, no doubt. Then you’ve probably got 6-8 weeks to use the service before they kick the bucket and shut down forever…
The APK is called “R1 Launcher”…
They’ve already proved it works from an Android phone.
