Update philosophy.py

philosophy.py

@@ -30,17 +30,27 @@ RATE_LIMIT = 3
 def create_service_context():
 
 #    prompt_helper = PromptHelper(max_input_size, num_outputs, max_chunk_overlap, chunk_size_limit=chunk_size_limit)
-    max_input_size = 4096
-    num_outputs = 512
-    max_chunk_overlap = 20
-    chunk_size_limit = 600 
-    prompt_helper = PromptHelper(max_input_size, num_outputs, chunk_overlap_ratio= 0.1, chunk_size_limit=chunk_size_limit)
-#    llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.7, model_name="gpt-4", max_tokens=num_outputs))
-
+#    max_input_size = 4096
+#    num_outputs = 512
+#    max_chunk_overlap = 20
+#    chunk_size_limit = 600 
+#    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-3.5-turbo", max_tokens=num_outputs))
+#    llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.5, model_name="gpt-3.5-turbo", max_tokens=num_outputs))
+
+    # Constraint parameters
+    max_input_size = 32768
+    num_outputs = 30000  # Safe upper limit to ensure total tokens do not exceed 32768
+    max_chunk_overlap = 20
+    chunk_size_limit = 600
+
+    # Allows the user to explicitly set certain constraint parameters
+    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-4-32k", max_tokens=num_outputs))
+    
     #constructs service_context
     service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor, prompt_helper=prompt_helper)
     return service_context