Had an interesting conversation on the science of consciousness a while back and realized there is a lot I don't know.

Can anyone recommend a recent book that covers the overall basics, such as the different theories (like antenna/etc) and physics overlaps?

My only requirement is scientific references. I dont want or need cited peer reviewed papers, but I don't want an author's sole opinion, either.

Infuriating GitHub users that mass-spam GitHub email addresses to promote their repo. How do I report this behavior to GitHub? I guess I could also report them for CANN Spam violations.

Tiny11 builds a Windows 11 ISO that is 3.7GB in size, and installs in less than 10GB. It is so stripped down it doesn't even have a web browser, so it requires minimal updates, and runs great as a VM.

Personally, I use it for things like configuring Webcams or Controllers.

I recommend installing in KVM using a qcow2 disk image named Tiny11-Base.qcow2, then:

Install VirtIO tools

https://fedorapeople.org/groups/virt/virtio-win/direct-downloads/archive-virtio/

To get VirtIO on there you can use a USB stick, or use PowerShell as Administrator to download:

$url = "https://fedorapeople.org/groups/virt/virtio-win/direct-downloads/archive-virtio/virtio-win-0.1.285-1/virtio-win-guest-tools.exe"
$output = "c:\Users\User\Download\virtio.exe"

Import-Module BitsTransfer
Start-BitsTransfer -Source $url -Destination $output

Mount folders

This is so you don't need a browser or to use PowerShell for downloading. I mount my local Downloads folder.

Guide: https://www.debugpoint.com/kvm-share-folder-windows-guest/

Latest FSP (Fuse for Windows) at time of writing: https://github.com/winfsp/winfsp/releases/download/v2.2B1/winfsp-2.2.26112.msi

Snapshot and backup

Shutdown and create backups.

This retains sparse file compatibility (a disk that can grow as needed).

• Snapshot will give you two files. Point your VM to the second one to use the snapshot.
• The backup command will backup the original and the snapshot.

Snapshot

qemu-img create -f qcow2 -b Tiny11-Base.qcow2 -F qcow2 Tiny11-Active.qcow2

Backup

tar --use-compress-program=pbzip2 -cSvf Tiny11.tar.bz2 Tiny11*.qcow2

Export KVM

virsh dumpxml Tiny11 > Tiny11.xml 

Restore qcow2

tar --use-compress-program=pbzip2 -xvf ./Tiny11.tar.bz2

Import KVM

virsh define Tiny11.xml 

I have no cat. But want month cat.

Nominate Broccoli Cheddar.

In case anyone is interested, I decided to test out LLama.cpp's new OpenVino backend to see how it compares on Intel GPUs. At first glance, it stomps all over the previous best-case, SYCL, but lags behind LLM-Scaler (Intel's VLLM fork), likely just due to the hardware optimizations against GPTQ/Int4. Interestingly tg512 was fastest on SYCL, but in real world, the prompt processing always seems the be the indicator on this card.

As usual with Intel, model selection is... poor. It took a while to even find a model that was in the validated OpenVino list that would not only run properly, but also have a counterpart that was "close enough" for LLM Scaler.

## Llama.cpp OpenVino
llama-benchy http://localhost:8000/v1 bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF:Q4_K_M

| model                                              |   test |              t/s |     peak t/s |      ttfr (ms) |   est_ppt (ms) |   e2e_ttft (ms) |
|:---------------------------------------------------|-------:|-----------------:|-------------:|---------------:|---------------:|----------------:|
| bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF:Q4_K_M | pp2048 | 3845.61 ± 524.73 |              | 659.99 ± 56.95 | 489.07 ± 56.95 |  739.42 ± 56.84 |
| bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF:Q4_K_M |  tg512 |     40.89 ± 0.55 | 44.33 ± 1.25 |                |                |                 |

## Llama.cpp SYCL
llama-benchy http://localhost:8000/v1 bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF:Q4_K_M

| model                                              |   test |            t/s |     peak t/s |       ttfr (ms) |    est_ppt (ms) |   e2e_ttft (ms) |
|:---------------------------------------------------|-------:|---------------:|-------------:|----------------:|----------------:|----------------:|
| bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF:Q4_K_M | pp2048 | 844.64 ± 19.25 |              | 2199.90 ± 23.63 | 2178.96 ± 23.63 | 2229.67 ± 24.84 |
| bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF:Q4_K_M |  tg512 |   73.87 ± 1.17 | 78.00 ± 2.16 |                 |                 |                 |

## LLM-Scaler
llama-benchy http://localhost:8000/v1 jakiAJK/DeepSeek-R1-Distill-Llama-8B_GPTQ-int4

| model                                             |   test |              t/s |     peak t/s |      ttfr (ms) |   est_ppt (ms) |   e2e_ttft (ms) |
|:--------------------------------------------------|-------:|-----------------:|-------------:|---------------:|---------------:|----------------:|
| jakiAJK/DeepSeek-R1-Distill-Llama-8B_GPTQ-int4    | pp2048 | 7875.52 ± 642.20 |              | 268.09 ± 20.50 | 240.11 ± 20.50 |  268.34 ± 20.45 |
| jakiAJK/DeepSeek-R1-Distill-Llama-8B_GPTQ-int4    |  tg512 |     52.75 ± 0.10 | 54.00 ± 0.00 |                |                |                 |

Pronounced "searching".

cross-posted from: https://lemmy.world/post/44792435

Hi,

I run a private, non-federated Matrix server (synapse) that I use as a way to combine my messaging services into one app (Element). I'm currently using Google Voice with the Mautrix-gvoice bridge, but I am fully de-Googling.

What would be the process for connecting multiple JMP.chat numbers in via a bridge? I assume by using the XMPP bridge in some way.

I plan to still use the Cheogram app for calling.

Thanks.

Hi,

I run a private, non-federated Matrix server (synapse) that I use as a way to combine my messaging services into one app (Element). I'm currently using Google Voice with the Mautrix-gvoice bridge, but I am fully de-Googling.

What would be the process for connecting multiple JMP.chat numbers in via a bridge? I assume by using the XMPP bridge in some way.

I plan to still use the Cheogram app for calling.

Thanks.

Hi all,

I'm looking for a GUI way to utilize a VPN over TLS. I can host whatever backend VPN is needed to tunnel through, Wireguard, OpenVPN, SSH, anything. My goal is to have the VPN travel over port 443 as TLS traffic and still be able to use the native android blocks for when the VPN is not active. I currently use Wireguard, but need to convert to a TLS solution.

So far the most promising and up-to-date I've found is WSTunnel: https://github.com/erebe/wstunnel with this guide from Hetzner: https://community.hetzner.com/tutorials/obfuscating-wireguard-using-wstunnel, however this requires me to run a command line app in Termux and then the WG client on top of that, vs having a full GUI option. I'll go down this path if I need to, but want to make sure I'm not missing something.

Thanks!

Hi all,

I'm doing an experiment with a cabinet for filament with a small dehumidifier in it. I can get it to sit at 35%, but not much lower. I've sealed the cabinet with foam tape, and dried most of the filament first.

Are there any tricks to help improve performance of the mini dehumidifier? Maybe adding more heat to the cabinet?

The goal would be to use Home Assistant and the humidity sensor to turn things on and off as needed.

Thanks!

Hi all,

American traveling to China for IETF, and making my tech prep plans (bringing a laptop, phone, tablet, kindle, and steam deck). I won't bias with my current plans too much, but I do already run Linux+LUKS and GrapheneOS.

For those with experience, what tech prep would you do?

Thanks!

I can't seem to locate one anywhere that isn't a 3-10 year old untouched repo on GitHub.

I've seen NFC Taginfo used a lot, but it appears to be proprietary. MIFARE Classic Tool is FOSS, but it's for more specific hardware than TagInfo.

Anyone have any suggestions?

I'm thinking of a gift for my partner for growing mushrooms at home for cooking (northeast US, near Canada).

Any suggestions on something easy to start with?

I've got an old nVidia Shield (2017) that doesn't handle 4K AV1. Unfortunately, it looks like AV1 is missing from the latest models, too.

Anyone have a good ad-free'ish suggestion?

The SV08 Max loses about 100mm of height when you add the enclosure, and while there have been some top hats designed before, they all look.. bad. All of them use clear plexiglass all the way around and look really weird and "hacked on".

So, I decided to design and make my own to match the factory enclosure as close as possible (and includes a Nevermore filter mount). Here's a preview, and as with all my projects I'll release I've released a write-up, directions, and links to the STEP/3MF files soon on my site: https://nowsci.com/mods-for-3d-printers/

Hi all,

The attached photo is my brim (top) and first layer (bottom) after stopping the print. I've noticed that my brims look totally fine, but the first layer of the print itself has gaps between the lines.

How do I go about debugging this? I'm using Orcaslicer with an SV08 Max, and customizations I have during slicing include dropping the nozzle .2mm with the following Machine gcode:

G90 ; Absolute
G21 ; Use metric
M83 ; Use relative distances for extrusion
SET_GCODE_OFFSET Z=-0.2 MOVE=0

And a flow ratio of .96 with this ABS, which I commonly use on the Voron as well.

Thoughts?