monkmartinez

Yep, and there are thousands of them around the $30 dollar mark. The case and power supply included!!!!! Governments (Local and State in the US) auction them off cheap if you really want to buy in bulk.

People have adapted all manner of thin, thick and virtual clients to run Klipper. I had it running on a VM in VirtualBox before I moved it to a Jetson Nano.

The shortage of Raspberry Pi's is going to be their undoing. Everyone in the circles I travel (3D printing, Hobby CNC, Self-hosting, cyberdeck, plotters, etc.) are looking for replacements. There are quite a few contenders, and more recently lots of folks are grabbing thin clients (then installing Linux).

My thought process; In a silo, the raspberry pi isn't all that spectacular and needs a lot of do-dads/thing-i-majigs for even basic functionality. Power supplies and real HDD/SSD's for Rpi's can be a pain point. There are cheaper and better alternatives imo. What makes the Raspberry Pi so cool is the community and ease of use due to said community effort. If that community coalesce around another platform, the rpi foundation will be in big trouble.

You can only make that determination when that "someone" utilizes the asset in some manner or converts it back to the original currency.

I totally agree... crazy times in the past few years. I have heard that some people were able to borrow against their "wealth" (stock holdings, crypto holdings, vested ownership shares) for homes, cars, boats and more since Covid. If their "wealth" suddenly evaporates in the form of losses, they find themselves on the wrong side of the trade. Super duper risky and the appetite for these "products" was immense from what I understand.

Not true. You would need to know the cost to build the house and/or the price paid for the house prior to this selling event. Once its been sold again, we can assess whether there has been a net reduction or addition to the supply of money/credit/debt.

He was right tho. There should have been a recession way before the Lehman and subsequent housing bust. History is repeating itself again with the Covid downturn. The fact that the FED/government came to the "rescue" and pushed the recession into the future is irrelevant. The reasons he was positing recessions were sound and they eventually all came to fruition.

I know being early and/or late is considered being wrong. The fact remains he is calling it like he sees it. A reversion to the mean during a mania is a ballsy thing to predict. If the fundamentals are fucked, it will eventually correct.

Roubini = After receiving a BA in political economics at Bocconi University, Milan and a doctorate in international economics at Harvard University, he became an academic at Yale and a visiting researcher/advisor at the International Monetary Fund (IMF), the Federal Reserve, World Bank, and Bank of Israel.

CZ (Binance CEO) = Zhao attended McGill University in Montreal, Quebec, where he majored in computer science.

I am not into judgements based on education alone... Especially considering the recent developments with the MIT grads at FTX. However, calling Roubini a "pop economist" is really not accurate to the slightest degree.

Bitcoin is designed to combat inflation? Please show me how it does this. I am very skeptical of this claim.

Being a firefighter, I am tangentially intimate[1] with construction methods for dwellings, be they commercial or private. I am fascinated with concrete and the myriad of ways it is formed. My nerdy side really wants to build a concrete 3D printer and print a compound for my family. My profession reminds me that building codes and laws are mostly written in blood and are codes/laws for a reason.

How could we speed up the process of adopting codes across the US? 1 hour to my south in Mexico, one can basically build anything they want in concrete and there really isn't a regulatory body that would stop them. The probabilities of earthquakes and hurricanes in that area are quite low, no really reason to build for snow loads, but definitely for wind. Yet, on the other side of that imaginary line, in the same desert with the same probabilities; Contractors must build houses that can withstand loads and forces that will likely not happen if the house was around for 200 years.

Todays homes are not generally built with concrete except for the slab on grade. They burn hotter and faster than anytime in the history of shelter.[2] This is mostly due to the contents we place inside, but a large part is the gluing of structural elements out of fractional lumber. We seal and insulate these newer houses up tight, because they have very little thermal mass, with petroleum based products that contribute to fire load as well.

Concrete would be a better choice for homes in the US, but its cost prohibitive due to labor and materials. I don't know for sure, but the price for building with concrete seems like it is mostly artificial. That is, using 8 x 8 blocks for walls seems excessive when compared to the layered 2 x 4 -> 1/2" OSB sheathing (only on corners) -> Foam/chickenwire -> Stucco sandwich. Why don't we see tilt-up concrete homes? Or columnar concrete homes? It all just moves so slowly...

[1] New phrase? I am not an architect or contractor, but I know my way around most building systems past and present to include mechanical, electrical and structural.

[2] https://www.nist.gov/fire

what do you see as the "revolutionary potential" in blockchain? Genuinely curious.

A few months from now will be the best time to buy. You can renegotiate the terms of the loan, but you can never renegotiate the price you pay!!!!

I disagree. Supply constraints were a figment of unsustainable demand. There was simply too much money sloshing around to satisfy the hedonistic human treadmill of MORE!

Reduce demand, broadly, as credit tightens with increased interest rates. Can't borrow at basically free money rates across the board anylonger.

Translates into, Less money for stock buybacks, less Yolo with stimy checks, left over money to yolo is also reduced == the market returns to mean. People feel less wealthy == slow purchases. Reduced purchases == Business struggle. Businesses lay people off == less demand. All of this == Less driving, less food.

Really very simple.

Do it.

Are there peer reviewed papers or something on Arvix that explicitly dive into the "beyond high school level of biology" where sex is not defined at the cellular level (wrt Autosomes)? I am not interested in debating the sociological version, to be perfectly honest.

Except both sides say the same things in different ways. The side that limits bigotry, racism, homophobia, and transphobia (as you say)... explicitly promotes bigotry against ideas they don't hold dear (alt-right, etc), racial prejudice, glorification of transgender and homosexual values.

I don't care what you practice, just don't shove it in my face and tell me to love it.

Why bother with nginx at all tho? Why take the chance of borking your ENTIRE system with software you aren't intimately familiar with? Dockerize that shit and move on. If you want to get into the weeds with some bare metal installs, have at it. There just isn't enough time for me to learn every sub-system in the entire software stack to be productive. I say, pick your poison and get after it!

I completely agree with you and will take it even further.

I don't even futz with nginx anymore. You don't need to do that with tunnels. This assumes you have created a tunnel with required domain name. Very, very easy to setup.

1. Create container 2. Open Cloudflare ZeroTrust 3. Create a hostname for the new service or app; heim.example.com -> http://localhost:port 4. Save it. 5. Done. -> access the service from http://heim.example.com from anywhere in the world 5. For extra security, register the app in zerotrust with an email verification policy with emails you trust or use a token, or what ever auth you like.

If I want to run a bunch services like apprise, n8n, strapi, jupyter notebooks, nocodb, appflowy and more... why wouldn't I run these in containers? Using Docker and the tooling around it makes it super easy. I don't spend weeks configuring every tiny detail just to get the darn things installed.

What you find annoying is empowering tons of people like me to run lot of software. I can try lots of software I wouldn't have bothered with due to ease of setup. I don't bork my host machine when I make mistakes. I just start delete the container and start over, with a few adjusted settings before build or compose steps.

1337 hackers can run circles around me for sure, but I have a freaking datacenter in 3 computers at my house. All of them are internet facing with Cloudflare tunnels pointed at the services I want to share or use away from home. It has never been this good.

I am very much wishing for the same. I have used T0_3B (https://github.com/bigscience-workshop/t-zero) on my M40 with 24GB of vRAM. It was fun to get it working, but its not remotely comparable to GPT-3 (instructGPT).

Just to get OTP up for inference would require a very large spend. To use GPT-NeoX (20B parameters) for inference requires 45GB of vRAM minimally. Its hard for me to imagine using a 170B model for fun somewhere unless one has a large GPU farm or lots of money.

I am not sure regarding that specific case. I can tell you this: You need a domain name with DNS setup to point at cloudflare for the tunnels to work. The domains can be at namecheap, DO, or anywhere where you can modify the DNS provider tags. At least one machine needs to run the tunneling service. I believe it should be the host that will serve the traffic.

I use cloudflare tunnels. No need to do anything to my router. I have 1 open publicly accessible port (80) exposed to the internet via Docker container. In order to reach that port, you have to traverse the tunnel with TLS and my home IP is never exposed and always updated except to Cloudflare. Literally one command and few configs to set up on Cloudflare to proxy traffic to my localhost and various subdomains.

You can have application level security enabled on Cloudflare's side for administrative interfaces, like portainer. That adds another level of security and there are a multitude of ways to verify (Email, token, one time pin, etc.). Super, super easy to set up.

I am a dolt when it comes to networking and don't trust myself to make the correct configurations. I no longer pay for a VPS for my simple blog. I don't want to spend my time becoming an expert Sys Admin/Network Admin... so I chose to go with a tunnel.

There is very little criticism of Obsidian here, but you definitely pointed a few out. I am a OneNote user and Obsidian doesn't really compare to OneNote. If you read a lot of papers, you can have OneNote OCR the text and make it searchable.

There it is! I am with you 100%. All of these massive plugin systems are a giant time suck. You spend soooo much time just setting things up only to forget the little stuff you need to do when something breaks. Every plugin I have installed requires some level of configuration. The more feature rich plugins (with no surprise there) are basically apps in and of themselves.

I keep hearing praise about this being a simple app, but I found it to be very fiddly if you want more than just markdown rendered in your editor. Lots and lots of time to get things working/configured.

Sync has been a massive PITA for me with Obsidian. Unless you are willing to pay, there is going to be pain there eventually. I had similar problems with OneNote, but those sync problems have mostly disappeared. Further, OneNote's handwriting experience is really good. It is very easy to export all notes from OneNote to Markdown with Pandoc, so while I may be "locked" in, I can "get out" if I want.

Edit: So thank you Obsidian for helping me convert my notes in OneNote to markdown to try you out! I now have a verified escape hatch if I should ever need it. However, I am not sticking around. Too much trouble for little stuff and OneNote is just a really good all arounder.

This plugin misses mark big time. I tried it and tried it again. Trying to love it... just couldn't compare with OneNote.

I would say even easier on prem as you don't need to wade 15 layers deep to do anything. Since I have moved to hosting my own stuff at my house, I have learned that connecting a monitor and keyboard to a 'sever' is awesome for productivity. I know where everything is, its fast as hell, and everything is locked down. Monitoring temps, adjusting and configuring hardware is just better in every imaginable way. Need more RAM, Storage, Compute? Slap those puppies in there and send it.

For home gamers like myself, it's has become a no brainer with advances in tunneling, docker, and cheap prices on Ebay.

Are you in the USA? I just did some googling on this as I was unaware of laws regarding the garnishment of wages. I have subordinates that have wages garnished for Child Support every paycheck. At least in my state, you can most certainly garnish wages/property for a judgement in your favor. It will probably take a bit more time, but I would stick it out to get my money back. If the contractor has nice Stihl saws, I would remand some $$$ to get one of them from him. Possibly even the truck as the contractor has probably depreciated the asset for tax reasons.

[0]https://www.azcourts.gov/selfservicecenter/Garnishment/Garni...

You don't need to do any port forwarding and/or router config if you don't mind using Cloudflare Tunnels. Seriously, these tunnels are a game changer when it comes to self hosting. I have one running on my main computer and another on a Jetson Nano being used as a webserver.

Is the H5 file type that much different than whatever the Pytorch versions are using?

The model is loaded from Huggingface during the instantiation of the stable diffusion class. It is loaded as an H5 file which I believe is unique to Keras[0]. I don't have any experience with Keras so I can't say if that is good or bad. I wanted to see where they were getting the weights as the blog post didn't demonstrate an explicit loading function/call like Pytorch.

Gonna run it and see... although I have like 40GB of stable diffusion weights on my computer now.

[0] https://github.com/keras-team/keras-cv/blob/master/keras_cv/...

Correct, MBP's can run stable diffusion and other ML workloads on non-nvidia hardware. I clearly see this becoming a trend. GPT-J, Neo and NeoX run really well on Colab TPU's, again these are not made by Nvidia.

Training is dominated by Nvidia, I will not question that as most papers I have seen say something similar. I will say that I do not believe training will always be dominated by Nvidia's datacenter options. Two things that will hasten the withdraw from Nvidia; Cuda and hardware advances around the motherboard (ASICs, RAM proximity, PCIe lanes, data transfer planes, etc).

Think about this... what if a company released an ML training/Inference ASIC that used regular DDR4/NVMe, performed like 4 x A100's and cost $8000? Would you be interested? I would! I don't think this is too far off, there has to be someone working on this outside of Google, Apple and Meta.