yeah that checks out, although looks like they do have an example for running models on a raspberry pi pico 2: https://docs.pytorch.org/executorch/main/pico2_tutorial.html. The list of embedded platforms this can run on is probably greater than the list of backends, it just wouldn't have acceleration.

It is fairly well established that neurons in these artificial neural networks are polysemantic and information is represented in directions in the activation embedding space rather than neurons independently representing information (which is why anthropic is doing things like training sparse autoencoders). I haven't read the paper in depth but it seems like it is based on a fundamental misunderstanding about neurons in ANNs vs the brain.

It may be that there are just a few misfirings that dramatically degrade the results. Strokes in the human brain causing a few neurons to die can have outsized affects in the body...

Very cool! I was wondering, is a separate model performing speech recognition for the voice demos such as the game? The FunctionGemma model card only seems to show text input/output.

Yes a separate model is performing ASR in this case. Gemma270m (base, function, and others) are not multimodal out of the box.

That being said if someone in the community wanted to use other encoders like siglip and plug them into Gemma270m to make it multimodal that'd be a great way to have fun over break and build up an AI Eegineer resume :)

There's a hidden equal sign, so if you copy-paste the code you get a different answer than if you just do it in your head.

I wonder if they also want to create more of a market for their products such as the DGX Spark.

that was quite the wild read, haha. Looks like they're going to have further meetings about the ducks but the conclusion of this meeting is that: "There was consensus among the Zoning Board of Appeals that the applicant has demonstrated emotional support ducks are necessary but did not come to a consensus on what would be reasonable regarding the number of ducks, the size and location of the enclosure, and conditions of approval."

It's an interesting area of research, there is even some evidence that language experienced in utero affects speech perception: https://doi.org/10.1111/apa.12098.

I was just thinking this. The wikipedia "signs of ai writing" that another commenter linked to (https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing) mentions LLMs overuse the 'rule of three' (e.g. natural imperfections, consistent lighting, and realistic proportions), haha.

Scalp-recorded EEG does not measure action potentials, it can only measure the graded potentials of basically one type of neuron (pyramidal cells) in the cortex, which is a really tiny percentage of both neurons and electrical activity in the brain. Additionally, there is also the various roles neurotransmitters play in the brain, etc., and glial cells seem to also play an important role. So, it’s definitely not the case that there aren’t any gaps that need new science, and even if there weren’t, it’s a pretty big stretch from there to decoding all brain activity solely through the electrical component.

You're probably right, but that doesn't mean GP is wrong, just that they need to state their thesis more carefully. There's more science to be done there, but there's no reason to believe that new fundamental laws of physics are required to explain the brain.