91 karma · joined October 27, 2023
Sorkin presented a theory that Nvidia bought HuggingFace to protect OpenAI from legal consequences. That a lawsuit determining accountability of actions by LLMs could threaten the AI financial network.
Surely they need GPU capacity and would need memory for those GPUs but OpenAI doesn't build GPUs or any hardware, right? So did they pay to keep the supply locked up, or do they have the ability to put that ram into use?
I briefly dabbled with Aider some months back but never got any real work done with it. Without installing each one of these new tools I'm having trouble grokking what is changing about them that moves the LLM-assisted software dev experience forward.
I have a fire tablet that I’ve tried that with, but for various reasons I prefer to have Linux on all the things. As a long time Android phone user Android still gives me an irrational ick, non-standard android even more so.
Ideally all my home devices would controlled and managed by the same underlying OS and tooling
I have to stop being such a prude, it just frustrates me that after so many years I can’t buy a cheap Linux tablet
I just want affordable, linux-powered displays in a slim tablet form factor. Performance and battery life are not a priority for my use case
Unfortunately I feel anything sold as a tablet comes with the assumption that it needs to compete with an iPad and be used for content consumption and gaming.
Ive seen raspberry pi kits sold that do just about everything else but this.
And I am guessing that a part of the reason for a lack of any such RaspPi tablets is that marketing such a tablet would come with the need to negatively differentiate it from any similarly priced android tablet.
However I can think of many use cases, mainly for folks in the maker space, that are not content consumption or gaming or long battery life. I am thinking of dashboards or smart home control panels.
Right now I have a few raspi4s mounted on the back of an official touchscreen encased in an adjustable plastic stand. Been working great for years, but the size is clunky and processing power is more than what I need it for, which is just displaying a web page with some information and buttons.
Would love a thin display to mount on a wall near a door or have others lying flat on a table next to a beside or couch. Basically always plugged in but with an included battery for the odd moment when I need to carry it somewhere.
So many other uses i could think of.
Ive looked at Amazon Fire tablets, but the locked-down android and really android of any kind is just not something I am interested in.
Ive seen raspberry pis used for just about everything else but not this
Was it being closed weight obvious to you from the article? Trying to understand why I was confused. Had not seen the "Flash" designation before
Also 30B models can beat a semi-recent 235B with just some additional training?
Their benchmark table shows it beating Qwen3-235B-A22B
Does "Flash" in the name of a Qwen model indicate a model-as-a-service and not open weights?
The benchmark table shows this Flash model beating their Qwen3-235B-A22B. I dont see how that is possible if it is a 30B-A3B model.
I don't see a mention of a parameter count anywhere in the article. Do you? This may not be an open weights model.
This article feels a bit deceptive
The benchmark table in their article shows Qwen3-Omni-Flash-2025-12-01 (and the previous Flash) as beating Qwen3-235B-A22B. How is that possible if this is only a 30B-A3B model? Also confusing how that comparison column starts out with one model but changes them as you descend down the table.
I don't see any FLASH variant listed on their Hugginface. Am i just missing it or are these specifying a model only used for their API service and there are no open weights to download?
Does anyone have insights on how compatible that hardware might be? Or how it might compare to something like a Pinetab?
For some years I have been looking for a low cost tablet with good Linux support for use in home automation or information displays. Surprised there is still nothing like this in the Raspberry Pi ecosystem.
Do users typically have one system-wide Jupyter install that gets reused for each data analysis project that then have their own dependencies in a virtual environment that Jupyter activates?
Or is Jupyter installed inside each project’s virtual environment?
I hope your grapes make it through
Being from the US, I don't recall any such stories in the news.
Back in 2022 there was a solar powered Airbus Zephyr drone that was tested over the Southwestern US with a flight time of 64 DAYS. I wonder how this new drone is different and how a 73 hour flight is significant in comparison.
Here is an article about the Zephyr Drone and its crash that ended its nearly record-tying flight:
https://simpleflying.com/airbus-zephyr-flight-ends/
Here is a flight replay from adsbexchange showing one day's worth of its flight path where it traced out the Liberty Bell(?) and the shape of the lower 48 at nearly 70,000ft. (Scrolling through its other dates show more playful flight paths)
https://globe.adsbexchange.com/?icao=ae1313&lat=33.419&lon=-...
Maybe the closed source UI is downloaded upon first execution for installation and then cached locally?
Or is this a web app that loads from the remote URL each time?
Stay in business, so monopoly arguments can be brushed aside.
But slowly erode privacy on the internet. And slowly lose user base.
their "AI" marketing page[1] pretty clearly implies that compute is offloaded
I think that answers most of my questions.I am also not in robotics, so this demo does seem quite impressive to me but I think they could have been more clear on exactly what technologies they are demonstrating. Overall still very cool.
Thanks for your reply
The article mentions that the system in each robot uses two ai models.
S2 is built on a 7B-parameter open-source, open-weight VLM pretrained on internet-scale data
and the other S1, an 80M parameter cross-attention encoder-decoder transformer, handles low-level [motor?] control.
It feels like although the article is quite openly technical they are leaving out the secret sauce? So they use an open source VLM to identify the objects on the counter. And another model to generate the mechanical motions of the robot.What part of this system understands 3 dimensional space of that kitchen?
How does the robot closest to the refrigerator know to pass the cookies to the robot on the left?
How is this kind of speech to text, visual identification, decision making, motor control, multi-robot coordination and navigation of 3d space possible locally?
Figure robots, each equipped with dual low-power-consumption embedded GPUs
Is anyone skeptical? How much of this is possible vs a staged tech demo to raise funding?Its a Debian Linux based e-ink tablet. It doesn’t look particularly premium, but it’s not android. There is a pressure sensitive stylus that comes with it as well.
https://pine64.org/devices/pinenote/
Not many reviews out there but here is a video of it:
Congrats, this looks nice
Does this app render the flights to fit the curvature of the earth?
Was looking for something like this during the New Jersey drone flap to help identify distant flights from the perspective of someone on the ground
The series is called the Gateway Experience. Its a kind of semi guided meditation using the binaural beats that is intended to induce out of body experiences.
I tried it years ago and although I never experienced a OOBE it did seem to induce lucid dreaming. Its a bit new-agey but not in a preachy or religious way.