Update src/RAGSample.py

src/RAGSample.py

@@ -391,10 +391,11 @@ Answer:
         "text-generation",
         model="microsoft/BioGPT",
         tokenizer="microsoft/BioGPT",
-        max_new_tokens=100,      # Reduced
-        max_length=400,          # Reduced
-        temperature=0.2,
-        device_map="cpu",        # Force CPU if GPU memory is limited
+        max_new_tokens=100,      # Reduced for stability
+        max_length=1024,         # BioGPT's context length
+        temperature=0.2,         # Lower for more focused responses
+        device_map="auto",
+        torch_dtype=torch.float16,
         return_full_text=False,
         truncation=True,
         do_sample=True,