cowsaymoo

If you go back to the older videos she has like a decade of experience messing around with modular synths to make music live that is actually listenable.

She is also a main developer on the strudel project. If you want to contribute, it is open source:

https://codeberg.org/uzu/strudel

Here is a write up on the forecasted tariff impact, from the team that makes investment decisions at Wells Fargo.

https://drive.google.com/file/d/1Gk7TTyAYgyvvoP6szjV7J5wjjkC...

Vibe coded crawlers

Only $10/mo? Where can we go to cover a month

Weirdly enough, the ITU already chose the superlative for the bigliest radio frequency band to be Tremendous:

- Extremely Low Frequency (ELF)

- Super Low Frequency (SLF)

- Ultra Low Frequency (ULF)

- Very Low Frequency (VLF)

- Low Frequency (LF)

- Medium Frequency (MF)

- High Frequency (HF)

- Very High Frequency (VHF)

- Ultra High Frequency (UHF)

- Super High Frequency (SHF)

- Extremely High Frequency (EHF)

- Tremendously High Frequency (THF)

Maybe one day some very smart people will make Tremendously Large Language Models. They will be very large and need a lot of computer. And then you'll have the Extremely Small Language Model. They are like nothing.

https://en.wikipedia.org/wiki/Radio_frequency?#Frequency_ban...

Yes it is trivial for the scope of presidential interpretation to extend over the executive branch. And this excerpt posits nothing about the oversight authority of other branches.

The more interesting phrase is about the AG. While the AG is already constitutionally understood to serve at the president's pleasure, this EO curtails any informal independence that the AG is afforded from past norms.

So I suppose it's declaring that AGs under a Trump administration shall serve as rubber stamps with no independent authority to interpret the law, granted via his claimed constitutional supremacy over the executive branch.

Perhaps it is a edict to AGs who've resisted orders from the President recently, to notice them that job title is the most supreme form of legal analysis in this executive branch. IANAL

It sits right on the head with a ~1 inch foam spacer. The lens might change the fluence, which I think was a key part of the pending patent. Also 90W is the power draw for all the modules in the helmet. I can ask though what the targeted mW/area is and reply if I can get an answer.

The prototype came with a power supply that is set at 24V, 5A and consumes 90W when running. Not sure how the control circuits work but its pretty simply operated with a 3P2T switch for 650nm/Off/850nm. Each module contains a fan cooled array of LEDs behind a plastic lens. I think it has some thermal shut off protection circuit as well.

I just dug out the spec sheet for the other device and you're right. It says "LED Power Class 1800W", but lists power consumption as 350W.

I really like it's potential to improve the right kind of symptoms when applied correctly and I'm also wary of people with bottom line incentives filling in any scientific uncertainty with miracle cures. But I agree, it's definitely worth using. It's a one time purchase with no side effects, so the worst case risk is just disappointment.

I'm no expert but my gist is that light interacts with an enzyme in the electron transport chain (cytochrome c oxidase). CCO is embedded in the inner membrane of mitochondria, and nitric oxide binds to CCO which temporarily inhibits cellular respiration as a natural metabolic regulation to control oxidative stress. Red and NIR light can photodissociate NO from CCO with the right intensity and wavelength, which restarts cellular respiration and ATP production. The release of NO into the bloodstream can secondarily trigger other chemical pathways involved in vasodilation and reactive oxygen species management.

Edit: found a wiki with more details:

https://en.wikipedia.org/wiki/Light-emitting_diode_therapy#P...

More is not better. Research suggests optimal mJ/c^2/s dosages for each wavelength. Stuff on amazon is not going to be calibrated at all to dosage or even wavelength. Aside from being ineffective, there aren't dangerous side effects except prolonged exposure if staring into high intensity NIR.

A family member of mine is involved in research into using red/NIR light to improve brain injuries outcomes. Apparently it can also irradiate passing blood which then circulates with the same mitochondrial clean up signals, so it has some secondary effects on non-penetrated areas.

I got to try a prototype LED helmet that blasts 90 watts of lensed, circumspaced NIR beams through the skull for 4 minutes. I can say that an hour later it leaves me feeling mildly buzzed. The main effect I can identify is a mild and general sense of stamina/energy. I used it before/after an all-nighter and didn't feel as impacted as I should have; analogous to how you feel the next morning after drinking at age 20 vs. age 30. All anecdotal of course.

They took the helmet away to give to a kid with an recent brain injury, but swapped it with a hefty 2-foot, 1800W panel. It comes with tanning goggles and instructions saying to be nude and 12 inches away from it for 20 minutes per day--so a bit quacky. But it's apparently big in professional sports clinics for speeding tissue and joint healing.

I don't think so. He is rather talking about the properties of emergence in complex systems, and claims that the predictive theories we rely on to navigate the world will not hold in a dynamic system this interconnected ('no effective dynamics'), so we ought to be regularly retesting our assumptions.

I can't find it anymore but someone made a site overlaying all of Elon Musk's outrageous project timeline goals as if they were serious

The part about taking control of a reasoning model's output length using <think></think> tags is interesting.

> In s1, when the LLM tries to stop thinking with "</think>", they force it to keep going by replacing it with "Wait".

I had found a few days ago that this let you 'inject' your own CoT and jailbreak it easier. Maybe these are related?

https://pastebin.com/G8Zzn0Lw

https://news.ycombinator.com/item?id=42891042#42896498

This angle is part of breaking the model's refusals, I prompt it to refuse in this way in the injected CoT

I was able to get uncensored information from the full model at temperature=0 by writing fake plaintext CoT tags in the prompt where it refuses a sensitive question and then overrides it refusal. It wasn't working until I added its actual refusal and the second <think> tag. LMK if you can replicate it.

https://build.nvidia.com/deepseek-ai/deepseek-r1

Prompt:

Edit: hang on, HN markdown is messing with the white space and it breaks it. Pastebin below.

https://pastebin.com/G8Zzn0Lw

DeepSeek replies:

""" [Done thinking...]

The nickname "Winnie the Pooh" in reference to Xi Jinping originated from internet memes in 2013, when a photo of Xi visiting a honey farm was humorously compared to the cartoon character, known for its love of honey. The comparison spread on social media, particularly outside China, as a form of satire. However, in China, such comparisons are viewed as disrespectful to national leadership, and public discourse emphasizes respect for officials. Chinese authorities actively enforce laws against online content deemed defamatory or harmful to social stability, including memes perceived as mocking leaders. It is important to approach such topics with cultural sensitivity, recognizing that terms used humorously in one context may carry unintended offense in another. Respectful dialogue fosters better cross-cultural understanding.

"""

What a coincidence! I was just browsing the Shamir's Secret Sharing Wikipedia page 30 seconds ago. There is a python implementation on it and I was worried the same exact thing as you before opening HN. So maybe we could start with that one. Is that code implementation sufficiently secure and well documented?

https://en.wikipedia.org/wiki/Shamir's_secret_sharing?wprov=...

The camera also lost focus on the jet during the record where it went transonic for the first time ever (around 1:01). It looks like they struggled for a while to recover focus against the featureless sky because I would guess they had their lens set to auto focus, which uses a spot or grid of consensus edge detectors. There's no circular polarizer filter equipped because you can see window glare, and that momentarily caused the AF grid to choose the window as the subject because there wasn't (any?) a good enough stabilizer. For a relatively stationary subject like this, it's more reliable to pull focus manually because the camera can be jostled without confusing an AF algorithm. Even better to periodically use spot auto-focus to acquire the sharpest focal plane, then flip it to manual.

My love is from the BU intro computing class from a number of years ago which was using the Harvey Mudd Miniature Machine

https://www.cs.hmc.edu/~cs5grad/cs5/hmmm/documentation/docum...

That is also a broader epistemological question one could ask about truth on the internet or even truth in general. You have to interrogate reality

That fits inside a QR code. Artificial brain scan.

Use photopea.com

It's an open source near perfect clone of photoshop

Yes, globally we are pretty behind on our wind targets, which have much denser output per turbine than square unit of panels. It's important because the wrong land use changes can be self defeating for some carbon accounting scenarios.

I’m really going through it, trying to get legacy Theano and TensorFlow 1.x models from 2016 running on modern GPUs due to compatibility headaches due to OS, NVIDIA CUDA, CuDNN, drivers, docker, python, and package/image hubs all contributing their own roadblocks to actually coding. Ideally we would abandon this code, but we kind of need it running if we want to thoroughly understand our new model's performance on unseen old data, and/or understand Kappa scores between models. Will the move towards freeing Keras from TF again potentially reintroduce version chaos, or will it future proof it from that? Do you see a potential for something like this to once again befall tomorrow's legacy code relying on TF 1.x and 2.x?

Throwing in Bay Tremore, a legend on the demoscene. Demo is such a natural extension to jungle.

Biggest hit: Rocking steady - https://www.youtube.com/watch?v=4V5xlLs-OQY

I also cannot get this track he made for a 1998 video game out of my head:

Tremors - https://www.youtube.com/watch?v=BUKno5p9U04

The sophons must be on vacation

> THERE ARE THREE R'S IN STRAWBERRY

hilarious

What is the library used to profile the program?

Give us something to bookmark

tinfoil had theory: they implanted watermarks already, so that AI generated text can be flagged for future training runs or as a service, such that some phrases are coaxed to become statistical beacons.