Add the model argument. (#471)

candle-examples/examples/ggml/main.rs

@@ -248,7 +248,7 @@ impl ModelWeights {
 struct Args {
     /// GGML file to load, typically a .bin file generated by the quantize command from llama.cpp
     #[arg(long)]
-    model: String,
+    model: Option<String>,
 
     /// The initial prompt.
     #[arg(long)]
@@ -283,12 +283,24 @@ impl Args {
         };
         Tokenizer::from_file(tokenizer_path).map_err(anyhow::Error::msg)
     }
+
+    fn model(&self) -> anyhow::Result<std::path::PathBuf> {
+        let model_path = match &self.model {
+            Some(config) => std::path::PathBuf::from(config),
+            None => {
+                let api = hf_hub::api::sync::Api::new()?;
+                let api = api.model("TheBloke/Llama-2-7B-GGML".to_string());
+                api.get("llama-2-7b.ggmlv3.q4_0.bin")?
+            }
+        };
+        Ok(model_path)
+    }
 }
 
 fn main() -> anyhow::Result<()> {
     let args = Args::parse();
 
-    let mut file = std::fs::File::open(&args.model)?;
+    let mut file = std::fs::File::open(&args.model()?)?;
     let start = std::time::Instant::now();
     let model = Content::read(&mut file)?;