fix: fix typo (#1870)

Co-authored-by: Kevin Lin <kevinlin@Kevins-MacBook-Pro.local>

letta/agent.py

@@ -481,7 +481,7 @@ def _get_ai_reply(
                 first_message=first_message,
                 # streaming
                 stream=stream,
-                stream_inferface=self.interface,
+                stream_interface=self.interface,
                 # putting inner thoughts in func args or not
                 inner_thoughts_in_kwargs_option=inner_thoughts_in_kwargs_option,
             )

letta/llm_api/llm_api_tools.py

@@ -115,7 +115,7 @@ def create(
     use_tool_naming: bool = True,
     # streaming?
     stream: bool = False,
-    stream_inferface: Optional[Union[AgentRefreshStreamingInterface, AgentChunkStreamingInterface]] = None,
+    stream_interface: Optional[Union[AgentRefreshStreamingInterface, AgentChunkStreamingInterface]] = None,
     # TODO move to llm_config?
     # if unspecified (None), default to something we've tested
     inner_thoughts_in_kwargs_option: OptionState = OptionState.DEFAULT,
@@ -149,28 +149,28 @@ def create(
 
         if stream:  # Client requested token streaming
             data.stream = True
-            assert isinstance(stream_inferface, AgentChunkStreamingInterface) or isinstance(
-                stream_inferface, AgentRefreshStreamingInterface
-            ), type(stream_inferface)
+            assert isinstance(stream_interface, AgentChunkStreamingInterface) or isinstance(
+                stream_interface, AgentRefreshStreamingInterface
+            ), type(stream_interface)
             response = openai_chat_completions_process_stream(
                 url=llm_config.model_endpoint,  # https://api.openai.com/v1 -> https://api.openai.com/v1/chat/completions
                 api_key=model_settings.openai_api_key,
                 chat_completion_request=data,
-                stream_inferface=stream_inferface,
+                stream_interface=stream_interface,
             )
         else:  # Client did not request token streaming (expect a blocking backend response)
             data.stream = False
-            if isinstance(stream_inferface, AgentChunkStreamingInterface):
-                stream_inferface.stream_start()
+            if isinstance(stream_interface, AgentChunkStreamingInterface):
+                stream_interface.stream_start()
             try:
                 response = openai_chat_completions_request(
                     url=llm_config.model_endpoint,  # https://api.openai.com/v1 -> https://api.openai.com/v1/chat/completions
                     api_key=model_settings.openai_api_key,
                     chat_completion_request=data,
                 )
             finally:
-                if isinstance(stream_inferface, AgentChunkStreamingInterface):
-                    stream_inferface.stream_end()
+                if isinstance(stream_interface, AgentChunkStreamingInterface):
+                    stream_interface.stream_end()
 
         if inner_thoughts_in_kwargs:
             response = unpack_all_inner_thoughts_from_kwargs(response=response, inner_thoughts_key=INNER_THOUGHTS_KWARG)
@@ -317,8 +317,8 @@ def create(
         # They mention that none of the messages can have names, but it seems to not error out (for now)
 
         data.stream = False
-        if isinstance(stream_inferface, AgentChunkStreamingInterface):
-            stream_inferface.stream_start()
+        if isinstance(stream_interface, AgentChunkStreamingInterface):
+            stream_interface.stream_start()
         try:
             # groq uses the openai chat completions API, so this component should be reusable
             assert model_settings.groq_api_key is not None, "Groq key is missing"
@@ -328,8 +328,8 @@ def create(
                 chat_completion_request=data,
             )
         finally:
-            if isinstance(stream_inferface, AgentChunkStreamingInterface):
-                stream_inferface.stream_end()
+            if isinstance(stream_interface, AgentChunkStreamingInterface):
+                stream_interface.stream_end()
 
         if inner_thoughts_in_kwargs:
             response = unpack_all_inner_thoughts_from_kwargs(response=response, inner_thoughts_key=INNER_THOUGHTS_KWARG)

letta/llm_api/openai.py

@@ -157,7 +157,7 @@ def openai_chat_completions_process_stream(
     url: str,
     api_key: str,
     chat_completion_request: ChatCompletionRequest,
-    stream_inferface: Optional[Union[AgentChunkStreamingInterface, AgentRefreshStreamingInterface]] = None,
+    stream_interface: Optional[Union[AgentChunkStreamingInterface, AgentRefreshStreamingInterface]] = None,
     create_message_id: bool = True,
     create_message_datetime: bool = True,
 ) -> ChatCompletionResponse:
@@ -167,7 +167,7 @@ def openai_chat_completions_process_stream(
     on the chunks received from the OpenAI-compatible server POST SSE response.
     """
     assert chat_completion_request.stream == True
-    assert stream_inferface is not None, "Required"
+    assert stream_interface is not None, "Required"
 
     # Count the prompt tokens
     # TODO move to post-request?
@@ -220,8 +220,8 @@ def openai_chat_completions_process_stream(
         ),
     )
 
-    if stream_inferface:
-        stream_inferface.stream_start()
+    if stream_interface:
+        stream_interface.stream_start()
 
     n_chunks = 0  # approx == n_tokens
     try:
@@ -230,17 +230,17 @@ def openai_chat_completions_process_stream(
         ):
             assert isinstance(chat_completion_chunk, ChatCompletionChunkResponse), type(chat_completion_chunk)
 
-            if stream_inferface:
-                if isinstance(stream_inferface, AgentChunkStreamingInterface):
-                    stream_inferface.process_chunk(
+            if stream_interface:
+                if isinstance(stream_interface, AgentChunkStreamingInterface):
+                    stream_interface.process_chunk(
                         chat_completion_chunk,
                         message_id=chat_completion_response.id if create_message_id else chat_completion_chunk.id,
                         message_date=chat_completion_response.created if create_message_datetime else chat_completion_chunk.created,
                     )
-                elif isinstance(stream_inferface, AgentRefreshStreamingInterface):
-                    stream_inferface.process_refresh(chat_completion_response)
+                elif isinstance(stream_interface, AgentRefreshStreamingInterface):
+                    stream_interface.process_refresh(chat_completion_response)
                 else:
-                    raise TypeError(stream_inferface)
+                    raise TypeError(stream_interface)
 
             if chunk_idx == 0:
                 # initialize the choice objects which we will increment with the deltas
@@ -314,13 +314,13 @@ def openai_chat_completions_process_stream(
             n_chunks += 1
 
     except Exception as e:
-        if stream_inferface:
-            stream_inferface.stream_end()
+        if stream_interface:
+            stream_interface.stream_end()
         print(f"Parsing ChatCompletion stream failed with error:\n{str(e)}")
         raise e
     finally:
-        if stream_inferface:
-            stream_inferface.stream_end()
+        if stream_interface:
+            stream_interface.stream_end()
 
     # make sure we didn't leave temp stuff in
     assert all([c.finish_reason != TEMP_STREAM_FINISH_REASON for c in chat_completion_response.choices])