docs: improve llama.cpp install instructions. (#720)

Does two things:

1. Makes homebrew install the default and points users to build
instructions for other platforms (note: our previous instructions were
also only for Mac - this PRs makes the wording more generalised)
2. moves from `-m` to `--hf-file` this would make sure that the models
are cached

---------

Co-authored-by: Julien Chaumond <[email protected]>

Author: Vaibhavs10

packages/tasks/src/local-apps.ts

@@ -48,16 +48,19 @@ function isGgufModel(model: ModelData) {
 
 const snippetLlamacpp = (model: ModelData): string[] => {
 	return [
-		`
-## Install and build llama.cpp with curl support
-git clone https://github.com/ggerganov/llama.cpp.git 
+		`## Install llama.cpp via brew
+brew install llama.cpp
+
+## or from source with curl support
+## see llama.cpp README for compilation flags to optimize for your hardware
+git clone https://github.com/ggerganov/llama.cpp
 cd llama.cpp
 LLAMA_CURL=1 make
 `,
 		`## Load and run the model
-./main \\
+llama \\
 	--hf-repo "${model.id}" \\
-	-m file.gguf \\
+	--hf-file file.gguf \\
 	-p "I believe the meaning of life is" \\
 	-n 128`,
 	];