Run llama.cpp with tea – without the installation pain!

[mfts]

First of all, on behalf of open-source developers and users, thank you so much for porting LLaMA to C++ ❤️

Running open-source made easy

At tea¹, we love open-source, so we packaged up llama.cpp and download the 7B model via torrents.
All you need to get started is

1. Install tea: sh <(curl https://tea.xyz)
2. Run llama.cpp -p "Getting paid to write open source can be accomplished in 3 simple steps:"

How it works

By running the package command llama.cpp -p "...", tea automagically downloads all required dependencies and runs llama.cpp for you. All dependencies get downloaded into an isolated ~/.tea directory, which doesn't mess with your system installations.

If you are curious, you can take a look at the package file for llama.cpp here.

Footnotes

1. tea is the next-generation cross-platform package manager from the creator of Homebrew. ↩

[Waloumi]

where do i run this? cmd? please...

[SolsticeProjekt]

Hi. Found you. Got this:

04/19 23:33:07 [NOTICE] Downloading 1 item(s)

04/19 23:33:07 [ERROR] Exception caught while loading DHT routing table from /home/bla/.cache/aria2/dht.dat
Exception: [DHTRoutingTableDeserializer.cc:83] errorCode=1 Failed to load DHT routing table from /home/bla/.cache/aria2/dht.dat

04/19 23:33:07 [NOTICE] IPv4 DHT: listening on UDP port 6997

04/19 23:33:07 [NOTICE] IPv4 BitTorrent: listening on TCP port 6957

04/19 23:33:07 [NOTICE] IPv6 BitTorrent: listening on TCP port 6957
[#7fd220 0B/0B CN:0 SD:0 DL:0B]
[#7fd220 0B/0B CN:0 SD:0 DL:0B]
[#7fd220 0B/0B CN:0 SD:0 DL:0B]

I don't know if it actually does something, though.

edit: CTRL+C, running the command again ... and the error message vanished.
It's still not actually doing anything, though, but i'll give it some time.

Any infos would be greatly appreciated.

[ampersanda]

I think there's a problem with the magnet url of the llama.cpp torrent