how to begin with windows release pack

[dengzheng-cloud]

Prerequisites

just download release win avx2 pack, and download int4 llama2 7b model via net.
got one question about how to run llama.cpp with model i downloaded.
do i need compile all project like i did in ubuntu, or i just need download visual studio( have no idea about)

Expected Behavior

is there any way i could just enter ./llama.cpp ./model_path in cmd then i can begin to chat with model.

Current Behavior

just extract whole pack file, use this for my uncle who has no computer science context, so better without compile, thx
i only know how to use it in ubuntu, if you know about windows release pack usage, please comment, thank you, sincerely.

[staviq]

You extract your download, and open terminal window from there, and from that point, you use it like on Linux.

Fastest way to open a terminal from a folder on windows, is to open the folder, and shift+right-click on empty space between icons, and you'll get "open power shell here" option in the context menu.

There is no llama.cpp program, you get couple of difrent example binaries.

Typically, only two of them will be useful for you: main.exe or server.exe. If you run any of those, like so main.exe -h you will get a help text describing all parameters.

Minimal command line invocation would be something like this:

main.exe -m "patch_to_model" -p "your prompt"

If you want chat, I recommend using the server program, as it gives you web browser UI.

server.exe -m "path_to_model"

[dengzheng-cloud]

appreciate it, model downloaded is a little bit older, got magic number error, i'll try another model.