bagatur comments

docs/docs/how_to/tool_calling.ipynb

@@ -98,7 +98,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -153,7 +153,7 @@
     "# | output: false\n",
     "# | echo: false\n",
     "\n",
-    "%pip install -qU langchain langchain_openai\n",
+    "%pip install -qU langchain langchain_openai openai\n",
     "\n",
     "import os\n",
     "from getpass import getpass\n",
@@ -167,7 +167,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -178,6 +178,8 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
+    "### Force model to select specific tools\n",
+    "\n",
     "We can also use the `tool_choice` parameter to ensure certain behavior. For example, we can force our tool to call the multiply tool by using the following code:"
    ]
   },
@@ -244,6 +246,38 @@
     "As we can see, even though the prompt didn't really suggest a tool call, our LLM made one since it was forced to do so. You can look at the docs for [`bind_tool`](https://api.python.langchain.com/en/latest/chat_models/langchain_openai.chat_models.base.BaseChatOpenAI.html#langchain_openai.chat_models.base.BaseChatOpenAI.bind_tools) to learn about all the ways to customize how your LLM selects tools."
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Disable parallel tool calling (OpenAI only)\n",
+    "\n",
+    "By default, OpenAI will run multiple tools (if selected) in parallel. For instance, if you ask it the weather of 3 cities and there is a ``get_weather`` function it will call the function 3 times. To ensure that only a single tool call is made, you can set the ``parallel_tool_calls`` parameter to False."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "{'tool_calls': [{'id': 'call_6blH9pWNr9PfZEcbwBGzRI1M',\n",
+       "   'function': {'arguments': '{\"a\":2,\"b\":2}', 'name': 'Add'},\n",
+       "   'type': 'function'}]}"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "llm_with_tools = llm.bind_tools(tools, parallel_tool_calls=False)\n",
+    "llm_with_tools.invoke(\"Please call the first tool two times\").additional_kwargs"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},

libs/partners/openai/langchain_openai/chat_models/base.py

@@ -817,7 +817,6 @@ class BaseChatOpenAI(BaseChatModel):
         tool_choice: Optional[
             Union[dict, str, Literal["auto", "none", "required", "any"], bool]
         ] = None,
-        parallel_tool_calls: Optional[bool] = None,
         **kwargs: Any,
     ) -> Runnable[LanguageModelInput, BaseMessage]:
         """Bind tool-like objects to this chat model.
@@ -839,15 +838,11 @@ class BaseChatOpenAI(BaseChatModel):
                 False: no effect;
                 or a dict of the form:
                 {"type": "function", "function": {"name": <<tool_name>>}}.
-            parallel_tool_calls: Whether to run tool calls in parallel
-                Can either be True or False
             **kwargs: Any additional parameters to pass to the
                 :class:`~langchain.runnable.Runnable` constructor.
         """
 
         formatted_tools = [convert_to_openai_tool(tool) for tool in tools]
-        if parallel_tool_calls:
-            kwargs["parallel_tool_calls"] = parallel_tool_calls
         if tool_choice:
             if isinstance(tool_choice, str):
                 # tool_choice is a tool/function name
@@ -1084,7 +1079,7 @@ class BaseChatOpenAI(BaseChatModel):
                     "schema must be specified when method is 'function_calling'. "
                     "Received None."
                 )
-            llm = self.bind_tools([schema], tool_choice=True)
+            llm = self.bind_tools([schema], tool_choice=True, parallel_tool_calls=False)
             if is_pydantic_schema:
                 output_parser: OutputParserLike = PydanticToolsParser(
                     tools=[schema], first_tool_only=True