Move more models to candle-transformers (#796)

* Move dinov2.

* Move efficientnet.

* Move the quantized llama model.

* Move segment-anything.

candle-examples/examples/quantized/main.rs

@@ -12,7 +12,7 @@ use candle::quantized::{ggml_file, gguf_file};
 use candle::{Device, Tensor};
 use candle_transformers::generation::LogitsProcessor;
 
-mod model;
+use candle_transformers::models::quantized_llama as model;
 use model::ModelWeights;
 
 const DEFAULT_PROMPT: &str = "My favorite theorem is ";