josh

neural decoding models @neurode_labs | obsessed with experience

[josh @j__vinny]

Properties of minds (non exhaustive):
- minds appear voluminous but are flat
- they do not occupy physical space, but they share a boundary with it
- at this boundary they may be read from and written to
- at this boundary they may meet
- minds cannot yet overlap

[josh @j__vinny]

@zetalyrae I was just reflecting on how much I like the emojis yesterday ☹️

[josh @j__vinny]

what gives people like this the arrogance to be like "obviously everything is going to be fine and I know better than the people building it and it's shameful that they would imply otherwise"

[josh @j__vinny]

can someone PLEASE help me understand why the company building the god machine being appropriately concerned about the potential negative consequences of it is a bad thing?! like why is the discourse about this ad just full time poasters being like "how dare ant do this"

[Daniel @growing_daniel]

Out of everything in that ad this part was exceptionally weird and sinister and makes me think someone in marketing at anthropic is a silent Pause AI supporter

[josh @j__vinny]

also calling a convolution "patchifying" is an ick

[josh @j__vinny]

avoiding the wasted forward pass through a big ema decoder by making the target stream shallow is an interesting idea. makes me wonder if the signal reconstruction loss can provide the anti-collapse pressure on its own though? then you wouldn't even need the ema updated tokeniser

[Brain.space @go_brain_space]

STST-JEPA brings JEPA style self supervised learning to EEG 🧠

Rather than reconstructing the full waveform or predicting deep encoder states, the model learns to predict masked tokenizer latent shallow-targets from an EMA tokenizer. A small reconstruction head keeps the learned representation tied to the original EEG signal. The model uses 30 second multichannel windows, spatiotemporal block masking, coordinate aware channel pooling for mixed montages, and a 24 layer Transformer encoder.

The architecture is only part of the story. STST-JEPA was pretrained on 47,703 sessions, including 22,588 sessions collected with our proprietary 117 channel EEG headset. Owning the hardware lets us build a large, consistent, high density EEG dataset beyond the scale of typical public datasets 🦉

Blog post: https://t.co/M6oFaUMigX
Arxiv: https://t.co/ETEL9KdruN

@RoySegal2 @YSvechinsky @TomerFekete

[josh @j__vinny]

@francoisfleuret https://t.co/l0pIGG3jdm

[Ilya Sutskever @ilyasut]

if you value intelligence above all other human qualities, you’re gonna have a bad time

[josh @j__vinny]

is this content too esoteric?

[blue @theproteindance]

What positive reason exists to make this. Please, I'm begging, what positive reason could there be for this to exist

[josh @j__vinny]

sometimes when dreaming, my phenomenal awareness becomes so saturated that I am unable to move or talk. its like the awareness of my paralysed body bleeds into my dream and manifests as an impossibly heavy blanket of light

minds are weird man

[albs— @albfresco]

don’t worry, there will be jobs in the future

the jobs in the future:

[josh @j__vinny]

@themeghamachine the main loss is character level, which to me says they are targeting verbatim decoding. if they only cared about semantic gists, they would’ve picked a loss that reflected that.

[max! @maxsloef]

every day, claude chooses a new piece from the Met to be my background. when fable arrived, it chose "Approaching Thunder Storm", saying:

"[this is] the particular human posture — calm, attentive, a little fatalistic — of facing something far too large to do anything about."

[josh @j__vinny]

@maxsloef this does kinda read like the dystopian trope of forcing a thrashing mind into a box...

[josh @j__vinny]

@francoisfleuret "computer turn your thinking power to max and then move the loss calculation to an affine space"

[josh @j__vinny]

@torchcompiled I mean it depends on the degree to which the embeddings represent the features of prose that help you identify AI'ness

[Ethan @torchcompiled]

New post: Pangram and other AI-text-detection overall serves a good purpose, but understated fallibility and misplaced usage leads to more chaos.

Here, I review research on AI-text-detection along with their claims and company positioning

[josh @j__vinny]

in this timeline we get to unironically say shit like "yeah turn up the AI's thinking power to max"

[josh @j__vinny]

anyone who thinks the 3d printing revolution never came hasn't worked at an actual hardware company

[François Fleuret @francoisfleuret]

Research circa 2026