Hacker Newsnew | past | comments | ask | show | jobs | submit | more sosuke's commentslogin

I’ve searched and searched for a way to turn off the extra processing. The only methods are third party camera apps or doing burst mode for every picture.

I preferred it when they saved the original and processed photos. Now there is no way to turn it off.

Worst part for me is when I get to see a perfect picture that gets processed a blink later and can never be recovered.


> Worst part for me is when I get to see a perfect picture that gets processed a blink later and can never be recovered.

This has happened to me way too many times. I take a picture, it looks great, then shortly after it gets processed so badly it's all blurred up with no return because of some forced long exposure low light compensation feature that cannot be disabled.


If you want to try out the “fix” goto camera settings and turn on “prioritize faster shooting” then in the camera app on the take picture button: tap, hold, drag left. It should stay white and play a shutter sound. There is a chance to do a similar action that just starts a recording but the button goes red for that.


Thank you! This fixes the after-shot blur. Now I just need to find a way to disable HDR/color enhancements without a 3rd party app and I'm perfectly set.


Halide. If you don't mind subscriptions or the lifetime $75 the app I found that did everything I wanted is Halide. The pain of the problem for me didn't cross the threshold for me to buy the app but the trial of it worked. Funny I have several dozen selfies of me testing it out compared to the Camera app with all the options toggled.


I can’t reply to dang’s comment with a Wikipedia article but I was laughing when the link they posted resulted in a toast notification.

https://i.postimg.cc/bvSk0fzj/IMG-7616.png


New option needed besides concise. Think quietly on your own, then answer concisely.

Sometimes I think I'd prefer it to "think" before answering anyhow. The immediate thinking out loud text can be irritating for some irrational reason.


Just telling the llm to preface it's conclusion with "Answer:" in a parseable way would be better.

You're still paying for those thinking tokens, or at the very least have to wait for them to be generated.


I can see the author's perspective I think but the argument comes out like a little rant at the world and how things are than any real objection.

Places say no pets, which keeps out 99% of the pets, but then a more responsible or at least more determined 1% came up with a way to excuse their pet presence. They "earn" the ability to take their pet into areas otherwise restricted. Is that a class system? I wouldn't relate the two since it isn't something assigned at birth. However I know I'm woefully ignorant on these matters.

Pets are responsibility. They are entirely dependent on the owner for continued existence. I do not agree with relating them to Adderall prescriptions in any way shape or form.

------

Or! Can this be related to the viewpoint of "It is better that ten guilty men go free than that one innocent man be convicted." and we just put up with all the bad actors for the good of the people who really need it. Given my bad experiences in medicine I'd side with this argument if nothing else.


> They are entirely dependent on the owner for continued existence.

They are dependent on someone but an owner moving to a new apartment could theoretically find another person to take care of them.


I’m not well read on LLMs in spite of using them daily. The increase in performance seems incremental. As if they added another 0 to the number it might only go up the same percentage in output.

So I assume that the number is just one facet of increasing output quality. Is that a safe assumption? Like throwing more energy at a problem to improve output it only goes so far.


You can improve results with cleaner datasets and if you prioritise a certain goal like conversation, code compleation or reasoning.


I'm reading the Textbooks Are All You Need paper, which goes into this idea. The result of that research was Phi 1, and eventually Phi 3 (released a few days ago).


There has been incremental progress for about 1 year from open weight models worse than GPT-3.5 to models in the area of GPT-4.

Same for inference speed/cost: many many incremental improvements within 1 year add up.


No ideas about side projects or anything "productive" but for a concrete example look at SillyTavern. Making fictional characters. Finding narratives, stories, role-play for tabletop games. You can even have group chats of AI characters interacting. No good use cases for profit but plenty right now for exploration and fun.


Shooting from the hip here I’d say $800 would be a good start for all new parts. You might get down to $500 with used parts.

The most expensive piece in either build is a video card. You want to be able to load the LLM file into your video card RAM.

I just got a 16gb new card for this. I can load up to 34b models on it but poorly. Anything 13b or less runs perfectly. A 12gb card would be able to run 7b models and with the right training I think 7b models can be awesome.

New 4060ti Used 3090


Hah. Damn our pedantic ways. I was about to come comment the same on brutalist. The author created something they are proud of but weren’t able to accurately describe.

Author; great work! Keep making things and you’ll figure the rest out along the way. No I wouldn’t buy or use it but I see the effort and potential you put in. Don’t stop.


Thanks!


I don't expect to buy another Chromebook after I found out mine had planned obsolescence. After a certain date it would no longer update. I didn't realize this would happen when I bought it so it came as a surprise. Still a perfectly fine machine otherwise I just have to make time to figure out how to install some Linux distro.


You could try ChromeOS Flex https://support.google.com/chromeosflex/answer/11552529 That's from a startup Google bought earlier.


Yes, always check when the device was released before buying a Chromebook. They get 10 years of updates after release, 8 years for models released before 2021.


Fines always feel meaningless. Who gets the money?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: