HNHacker News
TopNewBestAskShowJobs

mlajtos

600 karma · joined January 28, 2021

https://mlajtos.mu
submissionscomments
mlajtos··on Show HN: A JavaScript function that looks and behaves like a pipe operator
Exactly right!

https://github.com/mlajtos/es1995

mlajtos··on If you're interested in eye-tracking, I'm interested in funding you
Eyefluence had an extremely impressive demo at AWE2016: https://youtu.be/iCZLll1l92g?t=480

> But there's way more to do. We've barely scratched the surface of what's possible with eye tracking and I'd love to take a second crack at it.

What do you have in mind? What would you like to see?

mlajtos··on Show HN: Khoj – Chat offline with your second brain using Llama 2
I am sorry.

Would some summary of previous day would be helpful to you? Is your memory problem only episodic, or does it extend to factual and kinesthetic as well?

mlajtos··on Show HN: Khoj – Chat offline with your second brain using Llama 2
Have anyone got something valuable from talking to your second brain? What kind of conversations are you trying to have?
mlajtos··on Show HN: An open-source, collaborative, WYSIWYG Markdown editor
> WYSIWYG is or is not compatible with markdow

Markdown keyboard shortcuts in WYSIWYG editor works well together.

mlajtos··on Neural networks in the 1990s
Yes, everything was hand coded (no autodiff) & Hinton loves Matlab.
mlajtos··on Understanding DeepMind's sorting algorithm
Just a wild guess – developers lack soft/managerial skills. (Overgeneralization.) In companies, this is accounted for because developers are shielded by managers. But in F/OSS you get to interact directly with the developers. As for the why devs lack soft skills – hold your hats – programming is basically commanding. Without "thank you", without "please" and with lot of swearing. No time for pleasant bullshit. (By this reasoning, devs using declarative/functional programming should be more polite than imperative programmers, right? :D)
mlajtos··on A developer's view of Vision Pro
Thank you very much for your replies. :)

4. So the battery can be charged while plugged into the headset. What happens when you pull the battery from headset? I am guessing insta black. Does it have some power-saving mode where only R1 is feeding images from cameras to displays without any computing possibilities?

mlajtos··on A developer's view of Vision Pro
1. How about drinking stuff with it? Is it possible without getting it off?

2. What about sleeping with it? VRC people do that, so they might be curious.

3. Is it usable outside?

4. What was the longest time you had it continuously on?

5. Have you ever dropped it? :D

mlajtos··on A developer's view of Vision Pro
The actual quote from Gruber was "There is no border in the field of vision — your field of view through Vision Pro exactly matches what you see through your eyes without it."

What does "border in the FoV" means?

mlajtos··on A developer's view of Vision Pro
There are infra-red illuminators.
mlajtos··on U-Net CNN in APL: Exploring Zero-Framework, Zero-Library Machine Learning
Yes, every ML grad student should write autodiff lib (ideally using dual numbers). But nobody should ever write backward pass for whole net. Researchers did that in 80s and 90s. Once we had autodiff, it became obvious that deriving backward pass by hand is no-go.
mlajtos··on U-Net CNN in APL: Exploring Zero-Framework, Zero-Library Machine Learning
I agree that implementing autodiff is nice exercise to grasp the whole stack. But its a ladder that has to be thrown away after you climbed it. In other words, write that autodiff for yourself, and then use production grade ML solution where you won't ever write backward pass by hand.

The APL paper doesn't have implementation of autodiff, rather 90s style hand derivations.

mlajtos··on U-Net CNN in APL: Exploring Zero-Framework, Zero-Library Machine Learning
Machine Learning. There are compilers like XLA, MLIR, TVM and others.
mlajtos··on U-Net CNN in APL: Exploring Zero-Framework, Zero-Library Machine Learning
Nobody should write backward pass by hand.
mlajtos··on Why Lisp Syntax Works
Reject PEMDAS, return to monky:

x * 2 + 1 - 3 % x

https://mlajtos.mu/posts/new-kind-of-paper-2

mlajtos··on PyTorch for WebGPU
Indeed. "Object-oriented" fluent notation is basically equivalent to infix notation.

https://mlajtos.mu/posts/new-kind-of-paper-2

mlajtos··on Apple’s new headset meets reality
Also, there is MetalFX for faster rendering time. Render at low-res, upscale to high-res.

https://developer.apple.com/documentation/metalfx

mlajtos··on User In Yer Face, a worst-practise UI experiment (2018)
There is whole subreddit dedicated to these kind of UIs: https://www.reddit.com/r/badUIbattles/top/?t=all

Enjoy.

mlajtos··on Petrucci Music Library
Indeed, this is not related to Dream Theater’s John Petrucci.
mlajtos··on BQNPAD
Brain Quotient Measure

Best Quality Marshall

Bizarrely Quirky Matrices

mlajtos··on PMC-LLaMA: Further Finetuning LLaMA on Medical Papers
Why do you think that the ratio of misdiagnosis would go up if there was a dialogue or a review process?

2. and 3. are interlinked. In my mind, medicine is the number one application for ML. Every doctor should have the best AI at their finger tips. One person cannot digest all the research that is being done; computer can. The benchmarks will go up.

mlajtos··on PMC-LLaMA: Further Finetuning LLaMA on Medical Papers
When there is a review process in place, it is usually followed. Mistakes should be expected in the tiniest commits and there should be processes that catch them. As you suggested, medical community does this ad-hoc and only for interesting/unsure cases and this is the result:

> The researchers estimate that 7.4 million misdiagnosis errors are made every year, 2.6 million people receive a harm that could have been prevented, and another 370,000 are permanently disabled or die because of the misdiagnosis.

Source: https://edition.cnn.com/2022/12/15/health/hospital-misdiagno...

mlajtos··on PMC-LLaMA: Further Finetuning LLaMA on Medical Papers
In software development sphere, there is a review process for all code that is written. No code gets into production without an explicit approval of some peer or supervisor.

When you get diagnosed by a doctor, usually no one checks the diagnosis. I would love if my health was checked by medical professional committee. Human and/or AI.

mlajtos··on Geoffrey Hinton leaves Google and warns of danger ahead
Charter – https://openai.com/charter

Usage Policies – https://openai.com/policies/usage-policies

Is there anything in particular that you disagree with or was it only general negative comment? What do you see as direct evidence of them not deploying (baby) AGI in a safe manner? Is this about "100x capped for profit" strategy?

mlajtos··on Geoffrey Hinton leaves Google and warns of danger ahead
You are partially right — OpenAI is way ahead of everybody else. Even though OpenAI team is thinking and doing everything for safe deployment of (baby) AGI, public and experts don’t think this should be effort lead by single company. So Google naturaly wants to be the counterweight. (Ironic that OpenAI was supposed to be counterweight, not vice versa.) However, when you want to catch up somebody, you cheat. And cheating with AI safety is inherenťy dangerous. Moratorium for research and deployment just doesn’t make sense from any standpoint IMO.

Regarding the hand-holding: As Hinton noted, simple extrapolation of current progress yields models that are super-human in any domain. Even if these models would not be able to access Internet, in wrong hands it could create disaster. Or even in good hands that just don’t anticipate some bad outcome. Tool that is too powerful and nobody tried it before.

mlajtos··on Improbably Right
Thank you :)
mlajtos··on Improbably Right
> fallible and happy to be corrected

Thank you, that is a very accurate description of how I operate. If you don't mind, I'll steal this phrase from you. :)

mlajtos··on Improbably Right
> I think it would come across better if the author mixed times where their non-traditional intuition gave them the wrong answer.

Oh, there are a lot of instances where I was biased and my actions were dead wrong. (example story: https://mlajtos.mu/posts/i-told-my-crush-her-leaked-password ) This post was about rare moments when I was right. I'll try to write about my failures more often, as the source material is basically endless.

mlajtos··on Improbably Right
Sapir-Whorf to the rescue! :D
← PreviousPage 4 of 13Next →