llama.cpp 2950 (New formula)

.github/autobump.txt

@@ -1344,6 +1344,7 @@ literate-git
 little-cms2
 livekit
 livekit-cli
+llama-cpp
 llm
 llvm
 lmdb

Formula/l/llama.cpp.rb

@@ -0,0 +1,41 @@
+class LlamaCpp < Formula
+  desc "LLM inference in C/C++"
+  homepage "https://github.com/ggerganov/llama.cpp"
+  # pull from git tag to get submodules
+  url "https://github.com/ggerganov/llama.cpp.git",
+      tag:      "b2963",
+      revision: "95fb0aefab568348da159efdd370e064d1b35f97"
+  license "MIT"
+
+  livecheck do
+    throttle 10
+  end
+
+  depends_on xcode: ["15.0", :build]
+  depends_on arch: :arm64
+  depends_on macos: :ventura
+  depends_on :macos
+  uses_from_macos "curl"
+
+  def install
+    system "make", "LLAMA_FATAL_WARNINGS=ON", "LLAMA_METAL_EMBED_LIBRARY=ON", "LLAMA_CURL=ON"
+
+    bin.install "./main" => "llama-cli"
+    bin.install "./server" => "llama-server"
+  end
+
+  test do
+    llama_cli_command = ["llama-cli",
+                         "--hf-repo",
+                         "ggml-org/tiny-llamas",
+                         "-m",
+                         "stories15M-q4_0.gguf",
+                         "-n",
+                         "400",
+                         "-p",
+                         "I",
+                         "-ngl",
+                         "0"].join(" ")
+    assert_includes shell_output(llama_cli_command), "<s>"
+  end
+end