add with_main_gpu to LlamaModelParams

This commit adds a `with_main_gpu` method to `LlamaModelParams` which
allows the main GPU to be set.

Signed-off-by: Daniel Bevenius <[email protected]>

Author: danbev

llama-cpp-2/src/model/params.rs

@@ -56,6 +56,13 @@ impl LlamaModelParams {
         self
     }
 
+    /// sets the main GPU
+    #[must_use]
+    pub fn with_main_gpu(mut self, main_gpu: i32) -> Self {
+        self.params.main_gpu = main_gpu;
+        self
+    }
+
     /// sets `vocab_only`
     #[must_use]
     pub fn with_vocab_only(mut self, vocab_only: bool) -> Self {