added param, set temp

app.py

@@ -73,10 +73,13 @@ def generate_chat(prompt: str, llamaindex_llm: LlamaIndexPaLM):
             "framework": "Llama-Index",
             "index_type": "VectorStoreIndex",
             "vector_store": "Pinecone",
-            "model_name": "models/text-bison-001",
-            "temperture": 0.7,
-            "top_k": 40,
-            "top_p": 0.95,
+            "vector_store_index": llamaindex_llm._index_name,
+            "vector_store_namespace": llamaindex_llm._index_namespace,
+            "model_name": llamaindex_llm.llm._model_name,
+            # "temperture": 0.7,
+            # "top_k": 40,
+            # "top_p": 0.95,
+            "custom_kwargs": llamaindex_llm.llm._model_kwargs,
         },
         start_time_ms=start_time_ms,
         end_time_ms=agent_end_time_ms,
@@ -108,7 +111,8 @@ def generate_chat(prompt: str, llamaindex_llm: LlamaIndexPaLM):
                     'category': genai.types.HarmCategory.HARM_CATEGORY_UNSPECIFIED,
                     'threshold': genai.types.HarmBlockThreshold.BLOCK_NONE,
                 },
-            ]
+            ],
+            temperature=0.9,
         )
         result = response.result
         success_flag = "success"

src/llamaindex_palm.py

@@ -65,12 +65,14 @@ class LlamaIndexPaLMText(CustomLLM, extra=Extra.allow):
     def __init__(
         self,
         model_name: str = 'models/text-bison-001',
+        model_kwargs: dict = {},
         context_window: int = 8196,
         num_output: int = 1024,
         **kwargs: Any,
     ) -> None:
         super().__init__(**kwargs)
         self._model_name = model_name
+        self._model_kwargs = model_kwargs
         self._context_window = context_window
         self._num_output = num_output
         
@@ -93,7 +95,8 @@ class LlamaIndexPaLMText(CustomLLM, extra=Extra.allow):
                         'category': genai.types.HarmCategory.HARM_CATEGORY_UNSPECIFIED,
                         'threshold': genai.types.HarmBlockThreshold.BLOCK_NONE,
                     },
-                ]
+                ],
+                **self._model_kwargs
             )
             logging.debug(f"response:\n{response}")
             return response.candidates[0]['output']
@@ -155,12 +158,15 @@ class LlamaIndexPaLM():
     
     def set_index_from_pinecone(
         self, 
-        index_name: str = 'experience'
+        index_name: str = os.getenv('PINECONE_INDEX', 'experience'),
+        index_namespace: str = os.getenv('PINECONE_NAMESPACE', 'main')
     ) -> None:
         # Pinecone VectorStore
         pinecone_index = pinecone.Index(index_name)
-        self.vector_store = PineconeVectorStore(pinecone_index=pinecone_index, add_sparse_vector=True)
+        self.vector_store = PineconeVectorStore(pinecone_index=pinecone_index, add_sparse_vector=True, namespace=index_namespace)
         self.pinecone_index = VectorStoreIndex.from_vector_store(self.vector_store, self.service_context)
+        self._index_name = index_name
+        self._index_namespace = index_namespace
         return None
     
     def generate_response(