iaw
iaw on Hacker News
I think they just explained how corrupt the prior mayor was, this coupled with the disclosure by Mamdani about every prior mayor hiding air safety data after 9/11 paint a damning picture of the priorities of prior NYC mayors in recent memory.
Absence of evidence is not evidence of absence. With the behaviors we know OpenAI engages in the accusations are wholly believable.
This is a take... With both the internet and industrial control infrastructure any failure modes were studied, documented, and corrected.
The incompetence/negligence argument about OpenAI is completely valid given their failure to demonstrate the basic capabilities needed to develop advanced AI without major preventable externalties.
I run 3bit GLM5.2 and full precision Qwen3.6-27B. GLM is much much closer to frontier models in it's breadth and ability to plan. If you just need to implement Python code from an existing plan Qwen is your choice but it has problem succeeding with more complex tasks that GLM5.2 does not.
As I type this my local GLM5.2 is troubleshooting bugs that Qwen would not be able to handle.
On a model of this size quantization has much less impact on quality of output. I'm running a 3bit version and find it comparable to sonnet, almost opus.
No. Unsloth has CPU offloading. It'll be slow but it'll work even with SSD offloading.
I have a friend who may be dying. It's not cancer but equally dangerous. They caught it early but they're powerless to do anything. We're just watching, waiting, and hoping.
vLLM supports "banning" certain tokens but I don't know if it can dynamically reduce them.
To my knowledge you can also "ban" with llama.cpp but it is passed in the API call rather than to the server at initialization.
It's remarkable how capable GLM 5.1 is, what's amazing is the recent development of Qwen 3.6 27B being close in real world performance.
Can you back that up with.... you know... evidence? "regulation bad" tends to be a political talking point more than a valid argument when you're this vague.
Edited: to add, this speech talks a lot about the reduction in research funding from the US government which arguably has nothing to do with the regulatory environment.
This is a weird take. I get the desire to house people but someone choosing to rent rather than sell a home they own is not the crux of the issue. When there are corporations keeping swaths of housing empty to raise rent rates the real issue is market manipulation not small participants.
Thank you for this comment, there is no way I could eloquently explain my read on the comment you're replying to the way you did.
7x fewer major and minor collisions!!!!*
* Compared to the estimated U.S. average.
They have a huge store of data on accidents in teslas per mile driven. Why don't they compare their actual data on accidents? Well, they would, but it probably is worse with FSD.
I wonder if that display is because they would give them out not why.
Not to be crude but everything they mentioned
inside the bag was from a serious relationship. I really wonder if the outside of the bag was for the less serious relationships that were still candidates.
Or was it common for soldiers to give out pieces of their uniform to people they just met out?
For those curious like I was :
U+1F99C is a parrot
>According to a press release from the airline, economy seats on the retrofitted planes will have "back support with a fixed recline design," which in simpler terms means the seat will not have the ability to recline.
>The Premium section at the front of these planes will have "ergonomically contoured seat cushions, reclining seat backs and a large headrest with four-way adjustment capability."
So.. poor people don't deserve ergonomic seats?
I know this wont go over well but I think it's important to remember: Steve Jobs did not treat his pancreatic cancer when he was first diagnosed and it was treatable, he instead sought holistic/alternative medicine treatments[1]. He leveraged his vast wealth to get a replacement liver [2] that at the time was known to only extend his life. That liver could have gone to someone that did not have such a negative prognosis and possibly saved their life for a lot longer than it did Steve. He used a chunk of his remaining time to work on his last yacht.
He was a great product person but I don't think that excuses all of the horrible things he did as a human.
[1] https://www.psychologytoday.com/us/blog/food-for-thought/201...
[2] https://www.cnn.com/2009/HEALTH/06/24/liver.transplant.prior...
They have a well documented and known human slave trade for their laborers in the kingdom.
Maybe the narrative changes when their approaches towards human rights does?
> claude doesn't really know anything other than it's training data
I've seen cases where Claude demonstrates novel behaviors or combines existing concepts in new ways based on my input. I don't think it's as simple as memorization anymore.
The potential of fully miniature vacuum tubes was never realized
My routine is:
Prompt 1: <define task> Do not write any code yet. Ask any questions you need for clarification now.
Prompt 2: <answer questions> Do not write any code yet. What additional questions do you have?
Reiterate until questions become unimportant.
Oh! To clarify. I use fluoride in the morning and hydroxyapatite at night.
I recently started using a nano-hydroxyapatite based toothpaste. It can't restore enamel but does better at remineralization than fluoride, hopefully it will be a good intermediate for me until something regenerative is available.
Not sure I'm following. Is a 10 degree F increase in average summer temperatures over 25 years and a notably disrupted ecosystem 'summer'?
Now that you point this out, since I started using Claude my RSI pain is virtually non-existent. There is so much boilerplate and repetitive work taken out when Claude can hit 90% of the mark.
Especially with very precise language. I've heard of people using speech to text to use it which opens up all sorts of accessibility windows.
My experience has been that Claude can layout a lot of things in minutes that would take me hours if not days. Often I can dictate the precise logic and then claude get's most of the way there, with a little prompting claude can usually get even further. The amount of work I can get done is much more substantial than it used to be.
I think there is a lot of reticence to adopt AI for coding but I'm seeing it as a step change for coding the way powerful calculators/workstation computers were for traditional engineering disciplines. The volume of work they were limited to using a slide rule was much lower than now with a computer.
I've been using Claude to do the things that are straightforward that I don't want to for about a month now. The power of these development techniques is no where near fully tapped yet from what I can see.
I'm using Claude for my side projects. The pricing tiers are Free (for Sonnet 3.5), $20/m for Opus 4, $100/m for max (5x usage limit as $20) and then a $200 tier above that at another 5x I believe.
The usage limits reset every 6 hours.