handle partially quantized models (#76)

* handle partially quantized models

- fix for #53 #71 #69 #74
- in order to test the models
	- I added a default prompt of an appropriate form
	- while working on the model configuration also added additional stop tokens (#74)
- fixed the repetitionPenalty code (#71)

Libraries/LLM/Llama.swift

@@ -194,6 +194,13 @@ public class LlamaModel: Module, LLMModel {
         let (out, cache) = model(inputs, cache: cache)
         return (lmHead(out), cache)
     }
+
+    public func sanitize(weights: [String: MLXArray]) -> [String: MLXArray] {
+        // Remove unused precomputed rotary freqs
+        weights.filter {
+            !$0.key.contains("self_attn.rotary_emb.inv_freq")
+        }
+    }
 }
 
 public struct LlamaConfiguration: Codable {