Update src/RAGSample.py

src/RAGSample.py

@@ -387,18 +387,17 @@ Answer:
     
     # Initialize a local Hugging Face model
     hf_pipeline = pipeline(
-            "text-generation",
-            model="m42-health/Llama3-Med42-8B",
-            tokenizer="m42-health/Llama3-Med42-8B",
-            max_new_tokens=150,
-            max_length=2048,      # Llama3 supports longer context
-            temperature=0.3,
-            device_map="auto",
-            return_full_text=False,
-            truncation=True,
-            do_sample=True,
-            pad_token_id=128001,  # Llama3 pad token
-            eos_token_id=128009,  # Llama3 EOS token
+        "text-generation", 
+        model="Narrativaai/BioGPT-Large-finetuned-chatdoctor",
+        tokenizer="Narrativaai/BioGPT-Large-finetuned-chatdoctor",
+        max_new_tokens=100,
+        max_length=512,
+        temperature=0.3,               # Lower temp for medical accuracy
+        device="cpu",
+        return_full_text=False,
+        truncation=True,
+        do_sample=True,
+        torch_dtype="float32",
         )
     
     # Wrap it in LangChain