TheHobbyist

joined 1 year ago
[–] [email protected] 11 points 2 days ago

I see, so there is indeed a broader context to the burning alone, it was also with additional verbal hatred and then possibly the location, and the overall intention. I think this makes it clearer. Thanks

[–] [email protected] 28 points 2 days ago (14 children)

Not familiar with the guy himself who maybe does deserve criticism and prison, but about the Quran burning, is it genuinely fair to sentence someone to prison for that? Is it equivalent to burning the cross? The Swedish flag? I might be mission a broader context, but I don't feel like someone burning my symbol or flag should be punished with prison. Am I alone? I would hate it, don't get me wrong, but I still feel it goes in freedom of expression.

[–] [email protected] 1 points 2 days ago* (last edited 2 days ago)

The proud dad's name ends with Unis and the kid remembers the X first digits, hence Unix, hence Linux!

[–] [email protected] 51 points 2 days ago

This. I will resume my recommendation of Bitwarden.

[–] [email protected] 3 points 2 days ago (1 children)

I didn't say it can't. But I'm not sure how well it is optimized for it. From my initial testing it queues queries and submits them one after another to the model, I have not seen it batch compute the queries, but maybe it's a setup thing on my side. vLLM on the other hand is designed specifically for the multi co current user use case and has multiple optimizations for it.

[–] [email protected] 22 points 2 days ago* (last edited 2 days ago) (5 children)

I run the Mistral-Nemo(12B) and Mistral-Small (22B) on my GPU and they are pretty code. As others have said, the GPU memory is one of the most limiting factors. 8B models are decent, 15-25B models are good and 70B+ models are excellent (solely based on my own experience). Go for q4_K models, as they will run many times faster than higher quantization with little performance degradation. They typically come in S (Small), M (Medium) and (Large) and take the largest which fits in your GPU memory. If you go below q4, you may see more severe and noticeable performance degradation.

If you need to serve only one user at the time, ollama +Webui works great. If you need multiple users at the same time, check out vLLM.

Edit: I'm simplifying it very much, but hopefully should it is simple and actionable as a starting point. I've also seen great stuff from Gemma2-27B

Edit2: added links

Edit3: a decent GPU regarding bang for buck IMO is the RTX 3060 with 12GB. It may be available on the used market for a decent price and offers a good amount of VRAM and GPU performance for the cost. I would like to propose AMD GPUs as they offer much more GPU mem for their price but they are not all as supported with ROCm and I'm not sure about the compatibility for these tools, so perhaps others can chime in.

Edit4: you can also use openwebui with vscode with the continue.dev extension such that you can have a copilot type LLM in your editor.

[–] [email protected] 16 points 3 days ago (3 children)

I wouldn't assume this is done with malice in mind, but maybe this is someone unaware of the importance of a formal license.

[–] [email protected] 14 points 4 days ago* (last edited 4 days ago) (5 children)

Indeed. They only have in their github page:

Terms of Use

Feel free to use these components in personal and commercial projects. However, while the tutorials and demos are available for your use as-is, they cannot be redistributed or resold. Let’s keep things fair and respect each other’s work.

[–] [email protected] 10 points 4 days ago* (last edited 4 days ago) (4 children)

I'm wondering, the integrated RAM like Intel did for Lunar Lake, could the same performance be achieved with the latest CAMM modules? The only real way to go integrated to get the most out of it is doing it with HBM, anything else seems like a bad trade-off.

So either you go HBM with real bandwidth and latency gains or CAMM with decent performance and upgradeable RAM sticks. But the on-chip ram like Intel did is neither providing the HBM performance nor the CAMM modularity.

[–] [email protected] 16 points 1 week ago

They used PimEyes, nothing new.

Of importance: they do not want to release the tool but use it as a way to raise awareness.

[–] [email protected] 13 points 1 week ago (1 children)

The whole talk is available here: https://www.youtube.com/watch?v=ZNK4aSv-krI

This specific one is at 39min.

30
submitted 1 month ago* (last edited 1 month ago) by [email protected] to c/[email protected]
 

Hi folks,

I have Alpine Linux installed in an encrypted LUKS partition. I came across this tutorial which shows how to setup a key in a USB drive and when the drive is inserted and the computer booted, the LUKS partition auto-unlocks with the key on the USB drive.

https://askubuntu.com/questions/1414617/configure-ubuntu-22-04-zfs-for-automatic-luks-unlock-on-boot-via-usb-drive

I would like to setup the same thing but I do not have Alpine linux installed on ZFS, so I'm looking for ways to adapt the instructions.

So far, what I've done is:

  1. I've setup the key on the usb stick and I can unlock the LUKS partition with that key.
  2. create a /etc/mkinitfs/features.d/usb-unlock.sh script with the following content:

(the echo to /dev/kmesg was to check whether the script did indeed run at boot by trying to print to the kernel messages but I can't find anything in the kernel messages).

#!/bin/sh

echo "usb-unlock script starting..." > /dev/kmsg

USB_MOUNT="/mnt/my-usb-key" # The USB stick mounting point
LUKS_KEY_FILE="awesome.key"  # The name of your keyfile on the USB stick

# Search for the USB stick with the key
for device in $(ls /dev/disk/by-uuid/*); do
    mount $device $USB_MOUNT 2>/dev/null
    if [ -f "$USB_MOUNT/$LUKS_KEY_FILE" ]; then
        # Unlock the LUKS partition
        cryptsetup luksOpen /dev/sda3 cryptroot \
            --key-file "$USB_MOUNT/$LUKS_KEY_FILE" && exit 0
    fi
    umount $USB_MOUNT
done
echo "No USB key found, falling back to password prompt." # this message never appears, despite not having found the key on the usb stick

echo "usb-unlock script ending." > /dev/kmsg
  1. I added usb-unlock to the features in mkinitfs.conf:
mytestalpine:~# cat /etc/mkinitfs/mkinitfs.conf 
features="ata base ide scsi usb virtio ext4 cryptsetup keymap usb-unlock"
  1. run mkinitfs to rebuild the initramfs. Then reboot to test the implementation, which was unsuccessful.

What am I missing / doing wrong? Thank you for your help!

Edit: forgot to add step 4

 

Hi folks,

I'm seeing there are multiple services which externalise the task of "identity provider" (e.g. login with Facebook, google or what not).

In my case, I am curious about Tailscale, a VPN service which allows one to chose an identity provider/SSO between Google, Microsoft, Github, Apple and OIDC.

How can I find out what data is actually communicates to the identity provider? Their task should simply be to decide whether I am who I claim to be, nothing more. But I'm guessing there may be some subtleties.

In the case of Tailscale, would the identity provider know where I'm trying to connect? Or more?

Answers and insights much appreciated! The topic does not seem to have much information online.

 

Yesterday, there was a live scheduled by Louis Grossman, titled "Addressing futo license drama! Let's see if I get fired...". I was unable to watch it live, but now the stream seems to be gone from YouTube.

Did it air and was later removed? Or did it never happen in the first place?

Here's the link to where it was meant to happen: https://www.youtube.com/watch?v=HTBYMobWQzk

Cheers

Edit: a new video was recently posted at the following link: https://www.youtube.com/watch?v=lCjy2CHP7zU

I do not know if this was the supposedly edited and reuploaded video or if this is unrelated.

 

Hi folks,

I seem to be having some internet connectivity issues lately and I would like to monitor my access to the internet. I have a homelab and was wondering whether someone had perhaps something like a docker container which pings a custom website every so often and plots a timescale of when the connection was successful and when it was not.

Or perhaps you have another suggestion? I know of dashboards like grafana but I don't know whether they can be configured to actually generate that data or whether they rely on a third party to feed them. Thanks!

 

Hi folks, I'm looking for a specific YouTube video which I watched around 5 months ago.

The gist of the video is that it was comparing the transcoding performance of an Intel iGPU when used natively, compared to when passed through to a VM. From what I recall there was a significant performance hit and it was around 50% or so (in terms of fps transcoding). I believe the test was performed on jellyfin. I don't remember whether it was using xcpng, proxmox or another OS. I don't remember which channel published this video nor when it was published, just that I watched it sometime between April and June this year.

Anyone recall or know what video I'm talking about? Possible keywords include: quicksync, passthrough, sriov, iommu, transcoding, iGPU, encoding.

Thank you in advance!

view more: next ›