Update philosophy.py

philosophy.py

@@ -1,6 +1,7 @@
+import transformers  # Added since we use transformers.pipeline below
 from transformers import AutoModelForCausalLM, AutoTokenizer
+from llama_index.llm_predictor import LLMPredictor  # Updated import path for LLMPredictor
 from llama_index import (
-    LLMPredictor,
     PromptHelper,
     StorageContext,
     ServiceContext,
@@ -33,7 +34,7 @@ def create_llm_pipeline():
         "text-generation",
         model=model,
         tokenizer=tokenizer,
-        device=DEVICE,
+        device=0 if DEVICE == "cuda" else -1,  # Use device index: 0 for GPU, -1 for CPU
         max_length=2048,
         do_sample=True,
         temperature=0.7,
@@ -121,4 +122,4 @@ if __name__ == "__main__":
     # Create initial index
     index = data_ingestion_indexing("books")
     # Launch the interface
-    iface.launch()
+    iface.launch()