Where hot potato license?

HOT POTATO LICENSE Version 2, September 2021 All rights reserved by the last person to commit a change to this repository, except for the right to commit changes to this repository, which is hereby granted to all inhabitants of the Milky Way Galaxy for the purpose of committing changes to this repository.

Don’t give them ideas

Glad i could help ;)

You can get different results, sometimes better sometimes worse, most of the time differently phrased (e.g. the gemma models by google like to do bulletlists and sometimes tell me where they got that information from). There are models specifically trained / finetuned for different tasks (mostly coding, but also writing stories, answering medical questions, telling me what is on a picture, speaking different languages, running on smaller / bigger hardware, etc.). Have a look at ollamas library of models which is outright tiny compared to e.g. huggingface.

Also, i don’t trust OpenAI and others to be confidential with company data or explicit code snippets from work i feed them.

If you’re lucky you just set it to the wrong version, mine uses 10.3.0 (see below).

I tried running the docker container first as well but gave up since there are seperate versions for cuda and rocm which comes packaged with this as well and therefor gets unnecessary big.

I am running it on Fedora natively. I installed it with the setup script from the top of the docs:

curl -fsSL https://ollama.com/install.sh | sh

After that i created a service file (also stated in the linked docs) so that it starts at boot time (so i can just boot my pc and forget it without needing to login).

The crucial part for the GPU in question (RX 6700XT) was this line under the [service] section:

Environment="HSA_OVERRIDE_GFX_VERSION=10.3.0"

As you stated, this sets the environment variable for rocm. Also to be able to reach it from outside of localhost (for my server):

Environment="OLLAMA_HOST=0.0.0.0"

I have my gaming pc running as ollama host when i need it (RX 6700XT with rocm doing the heavy lifting). PC idles at ~50W and draws up to 200W when generating an answer. It is plenty fast though.

My mini pc home server is running openwebui with access to this “ollama instance” but also OpenAIs api when i just need a quick answer and therefor don’t turn on my pc.

It’s a reference to the latest fireship video… I think:

https://youtu.be/AdygBbbEnco

Wouldn’t want to write a webserver / database connection / scheduler / etc. from scratch. Spring Boot plus lombok turns 2k lines of code into 100.

I hope it went well :) i was completely ready to go back changing the image tag to v2 but didn’t need to.

Alpine

I like the idea of it and there were times i used it correctly, but most of the time i do it wrong i guess.

Bricked my pc twice because of the bootloader and couldn’t repair it. From now on i just nuke my system if something is fucky and have a shell script do the installing of packages etc.

I had a friend come over to my place to fix her laptops wifi. After about an hour searching for any setting in windows that i could have missed, i coincidentally found a forum where one pointed out this could be due to a hardware wifi switch…

Im so looking forward to this. When i tried to use tmpfs / ramdisk, the transcoding would simply stop because there was no space left.

Yes, since we have similar gpus you could try the following to run it in a docker container on linux, taken from here and slightly modified:

#!/bin/bash

model=microsoft/phi-2
# share a volume with the Docker container to avoid downloading weights every run
volume=<path-to-your-data-directory>/data

docker run -e HSA_OVERRIDE_GFX_VERSION=10.3.0 -e PYTORCH_ROCM_ARCH="gfx1031" --device /dev/kfd --device /dev/dri --shm-size 1g -p 8080:80 -v $volume:/data ghcr.io/huggingface/text-generation-inference:1.4-rocm --model-id $model

Note how the rocm version has a different tag and that you need to mount your gpu device into the container. The two environment variables are specific to my (any maybe yours also) gpu architecture. It will need a while to download though.

Huggingface TGI is just a piece of software handling the models, like gpt4all. Here is a list of models officially supported by TGI, although they state that you can try different ones as well. You follow the link and look for the files section. The size of the model files (safetensors or pickele binaries) gives a good estimate of how much vram you will need. Sadly this is more than most consumer graphics cards have except for santacoder and microsoft phi.

I tried Huggingface TGI yesterday, but all of the reasonable models need at least 16 gigs of vram. The only model i got working (on a desktop machine with a amd 6700xt gpu) was microsoft phi-2.

Meanwhile 4tb sata ssd is 300€ in germany

It’s true that you shouldn’t open ports to the internet. If you still want your services to be accessible from outside the local network you can install a wireguard server on your thin client that has access to the services you want. And if you really want to harden it you can restrict wireguard clients from ssh and other admin things.

You will need to open one port on the router to your wireguard server though. Wireguard is UDP though and ignores packages without an established connection, so attackers will not even know there is an open port on your router.

Edit: tailscale and zerotier are good external solutions to this as well without needing to open a port at all.

Wow, thats just plain stupid. I hope someone forked it.