Update philosophy.py

philosophy.py

@@ -43,8 +43,11 @@ def create_service_context():
 #    max_input_size = 32768          # gpt-4-32k
 #    num_outputs = 30000             # gpt-4-32k
 #    num_outputs = 3500 # max 4096   # gpt-4o
-    max_input_size = 8192
-    num_outputs = 7500  # Safe upper limit to ensure total tokens do not exceed 8192
+#    max_input_size = 8192 # gpt-4
+#    num_outputs = 7500  # gpt-4
+    max_input_size = 16384
+    num_outputs = 15000  # Safe upper limit to ensure total tokens do not exceed 8192    
+    
     max_chunk_overlap = 20
     chunk_size_limit = 600
 
@@ -52,7 +55,7 @@ def create_service_context():
     prompt_helper = PromptHelper(max_input_size, num_outputs, chunk_overlap_ratio=0.1, chunk_size_limit=chunk_size_limit)
 
     # LLMPredictor is a wrapper class around LangChain's LLMChain that allows easy integration into LlamaIndex
-    llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.5, model_name="gpt-4", max_tokens=num_outputs)) # gpt-4-32k
+    llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.5, model_name="gpt-3.5-turbo", max_tokens=num_outputs)) # gpt-4-32k
     
     #constructs service_context
     service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor, prompt_helper=prompt_helper)