"For basic creation and usage of a tool-calling ReAct-style agent, the functionality is the same. First, let's define a model and tool(s), then we'll use those to create an agent."
]
},
{
"cell_type": "code",
"execution_count": 6,
"execution_count": 1,
"id": "1e425fea-2796-4b99-bee6-9a6ffe73f756",
"metadata": {},
"outputs": [],
@ -72,7 +72,7 @@
},
{
"cell_type": "code",
"execution_count": 15,
"execution_count": 2,
"id": "03ea357c-9c36-4464-b2cc-27bd150e1554",
"metadata": {},
"outputs": [
@ -83,7 +83,7 @@
" 'output': 'The value of `magic_function(3)` is 5.'}"
]
},
"execution_count": 15,
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
@ -119,7 +119,7 @@
},
{
"cell_type": "code",
"execution_count": 16,
"execution_count": 3,
"id": "53a3737a-d167-4255-89bf-20ac37f89a3e",
"metadata": {},
"outputs": [
@ -130,7 +130,7 @@
" 'output': 'The value of `magic_function(3)` is 5.'}"
]
},
"execution_count": 16,
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
@ -150,7 +150,7 @@
},
{
"cell_type": "code",
"execution_count": 17,
"execution_count": 4,
"id": "74ecebe3-512e-409c-a661-bdd5b0a2b782",
"metadata": {},
"outputs": [
@ -158,10 +158,10 @@
"data": {
"text/plain": [
"{'input': 'Pardon?',\n",
" 'output': 'The result of applying the `magic_function` to the input `3` is `5`.'}"
" 'output': 'The result of applying `magic_function` to the input 3 is 5.'}"
]
},
"execution_count": 17,
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
@ -200,7 +200,7 @@
},
{
"cell_type": "code",
"execution_count": 18,
"execution_count": 5,
"id": "a9a11ccd-75e2-4c11-844d-a34870b0ff91",
"metadata": {},
"outputs": [
@ -211,7 +211,7 @@
" 'output': 'El valor de `magic_function(3)` es 5.'}"
]
},
"execution_count": 18,
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
@ -243,7 +243,7 @@
},
{
"cell_type": "code",
"execution_count": 14,
"execution_count": 6,
"id": "a9486805-676a-4d19-a5c4-08b41b172989",
"metadata": {},
"outputs": [],
@ -272,20 +272,16 @@
},
{
"cell_type": "code",
"execution_count": 21,
"execution_count": 7,
"id": "d369ab45-0c82-45f4-9d3e-8efb8dd47e2c",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'input': 'what is the value of magic_function(3)?',\n",
" 'output': 'El valor de magic_function(3) es 5. ¡Pandamonium!'}"
]
},
"execution_count": 21,
"metadata": {},
"output_type": "execute_result"
"name": "stdout",
"output_type": "stream",
"text": [
"{'input': 'what is the value of magic_function(3)?', 'output': 'El valor de magic_function(3) es 5. ¡Pandamonium!'}\n"
"With LangChain's [AgentExecutor](https://api.python.langchain.com/en/latest/agents/langchain.agents.agent.AgentExecutor.html#langchain.agents.agent.AgentExecutor.iter), you could add chat [Memory](https://api.python.langchain.com/en/latest/agents/langchain.agents.agent.AgentExecutor.html#langchain.agents.agent.AgentExecutor.memory) so it can engage in a multi-turn conversation."
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "1fb52a2c",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Hi Polly! The output of the magic function for the input 3 is 5.\n",
"---\n",
"Yes, I remember your name, Polly! How can I assist you further?\n",
"---\n",
"The output of the magic function for the input 3 is 5.\n"
" {\"input\": \"Hi, I'm polly! What's the output of magic_function of 3?\"}, config\n",
" )[\"output\"]\n",
")\n",
"print(\"---\")\n",
"print(agent_with_chat_history.invoke({\"input\": \"Remember my name?\"}, config)[\"output\"])\n",
"print(\"---\")\n",
"print(\n",
" agent_with_chat_history.invoke({\"input\": \"what was that output again?\"}, config)[\n",
" \"output\"\n",
" ]\n",
")"
]
},
{
"cell_type": "markdown",
"id": "c2a5a32f",
"metadata": {},
"source": [
"#### In LangGraph\n",
"\n",
"Memory is just [persistence](https://langchain-ai.github.io/langgraph/how-tos/persistence/), aka [checkpointing](https://langchain-ai.github.io/langgraph/reference/checkpoints/).\n",
"\n",
"Add a `checkpointer` to the agent and you get chat memory for free."
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "035e1253",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Hi Polly! The output of the magic_function for the input 3 is 5.\n",
"---\n",
"Yes, your name is Polly!\n",
"---\n",
"The output of the magic_function for the input 3 was 5.\n"
" (\"user\", \"Hi, I'm polly! What's the output of magic_function of 3?\")\n",
" ]\n",
" },\n",
" config,\n",
" )[\"messages\"][-1].content\n",
")\n",
"print(\"---\")\n",
"print(\n",
" app.invoke({\"messages\": [(\"user\", \"Remember my name?\")]}, config)[\"messages\"][\n",
" -1\n",
" ].content\n",
")\n",
"print(\"---\")\n",
"print(\n",
" app.invoke({\"messages\": [(\"user\", \"what was that output again?\")]}, config)[\n",
" \"messages\"\n",
" ][-1].content\n",
")"
]
},
{
"cell_type": "markdown",
"id": "d7cf24a8",
"metadata": {},
"source": [
"## Iterating through steps\n",
"\n",
"With LangChain's [AgentExecutor](https://api.python.langchain.com/en/latest/agents/langchain.agents.agent.AgentExecutor.html#langchain.agents.agent.AgentExecutor.iter), you could iterate over the steps using the [stream](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.Runnable.html#langchain_core.runnables.base.Runnable.stream) (or async `astream`) methods or the [iter](https://api.python.langchain.com/en/latest/agents/langchain.agents.agent.AgentExecutor.html#langchain.agents.agent.AgentExecutor.iter) method. LangGraph supports stepwise iteration using [stream](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.Runnable.html#langchain_core.runnables.base.Runnable.stream) "
"for step in agent_executor.stream({\"input\": query}):\n",
" print(step)"
]
},
{
"cell_type": "markdown",
"id": "46ccbcbf",
"metadata": {},
"source": [
"#### In LangGraph\n",
"\n",
"In LangGraph, things are handled natively using [stream](https://langchain-ai.github.io/langgraph/reference/graphs/#langgraph.graph.graph.CompiledGraph.stream) or the asynchronous `astream` method."
" AIMessage(content='The value of `magic_function(3)` is 5.', response_metadata={'token_usage': {'completion_tokens': 14, 'prompt_tokens': 87, 'total_tokens': 101}, 'model_name': 'gpt-4o', 'system_fingerprint': 'fp_729ea513f7', 'finish_reason': 'stop', 'logprobs': None}, id='run-e5d94c54-d9f4-45cd-be8e-a9101a8d88d6-0')]}"
]
},
"execution_count": 23,
"execution_count": 13,
"metadata": {},
"output_type": "execute_result"
}
@ -400,7 +664,7 @@
},
{
"cell_type": "code",
"execution_count": 24,
"execution_count": 14,
"id": "16f189a7-fc78-4cb5-aa16-a94ca06401a6",
"metadata": {},
"outputs": [],
@ -416,7 +680,7 @@
},
{
"cell_type": "code",
"execution_count": 26,
"execution_count": 15,
"id": "c96aefd7-6f6e-4670-aca6-1ac3d4e7871f",
"metadata": {},
"outputs": [
@ -431,7 +695,11 @@
"Invoking: `magic_function` with `{'input': '3'}`\n",
"\n",
"\n",
"\u001b[0m\u001b[36;1m\u001b[1;3mSorry, there was an error. Please try again.\u001b[0m\u001b[32;1m\u001b[1;3mParece que hubo un error al intentar obtener el valor de `magic_function(3)`. ¿Te gustaría que lo intente de nuevo?\u001b[0m\n",
"\u001b[0m\u001b[36;1m\u001b[1;3mSorry, there was an error. Please try again.\u001b[0m\u001b[32;1m\u001b[1;3m\n",
"Invoking: `magic_function` with `{'input': '3'}`\n",
"responded: Parece que hubo un error al intentar obtener el valor de `magic_function(3)`. Permíteme intentarlo de nuevo.\n",
"\n",
"\u001b[0m\u001b[36;1m\u001b[1;3mSorry, there was an error. Please try again.\u001b[0m\u001b[32;1m\u001b[1;3mAún no puedo obtener el valor de `magic_function(3)`. ¿Hay algo más en lo que pueda ayudarte?\u001b[0m\n",
"\n",
"\u001b[1m> Finished chain.\u001b[0m\n"
]
@ -440,10 +708,10 @@
"data": {
"text/plain": [
"{'input': 'what is the value of magic_function(3)?',\n",
" 'output': 'Parece que hubo un error al intentar obtener el valor de `magic_function(3)`. ¿Te gustaría que lo intente de nuevo?'}"
" 'output': 'Aún no puedo obtener el valor de `magic_function(3)`. ¿Hay algo más en lo que pueda ayudarte?'}"
]
},
"execution_count": 26,
"execution_count": 15,
"metadata": {},
"output_type": "execute_result"
}
@ -471,7 +739,7 @@
},
{
"cell_type": "code",
"execution_count": 29,
"execution_count": 16,
"id": "b974a91f-6ae8-4644-83d9-73666258a6db",
"metadata": {},
"outputs": [
@ -480,14 +748,11 @@
"output_type": "stream",
"text": [
"('human', 'what is the value of magic_function(3)?')\n",
"content='Sorry, there was an error. Please try again.' name='magic_function' id='06f997fd-5309-4d56-afa3-2fe8cbf0d04f' tool_call_id='call_9fMkSAUGRa2BsADwF32ct1m1'\n",
"content='Sorry, there was an error. Please try again.' name='magic_function' id='a931dd6e-2ed7-42ea-a58c-5ffb4041d7c9' tool_call_id='call_Fg92zoL8oS5q6im2jR1INRvH'\n",
"content='It seems there is an issue with processing the request for the value of `magic_function(3)`. Let me try a different approach.' additional_kwargs={'tool_calls': [{'id': 'call_lbYBMptprZ6HMqNiTvoqhmwP', 'function': {'arguments': '{\"input\":\"3\"}', 'name': 'magic_function'}, 'type': 'function'}]} response_metadata={'token_usage': {'completion_tokens': 43, 'prompt_tokens': 130, 'total_tokens': 173}, 'model_name': 'gpt-4o', 'system_fingerprint': 'fp_729ea513f7', 'finish_reason': 'tool_calls', 'logprobs': None} id='run-2e0baab0-c4c1-42e8-b49d-a2704ae977c0-0' tool_calls=[{'name': 'magic_function', 'args': {'input': '3'}, 'id': 'call_lbYBMptprZ6HMqNiTvoqhmwP'}]\n",
"content='Sorry, there was an error. Please try again.' name='magic_function' id='9957435a-5de3-4662-b23c-abfa31e71208' tool_call_id='call_lbYBMptprZ6HMqNiTvoqhmwP'\n",
"content='It appears that the `magic_function` is currently experiencing issues when attempting to process the input \"3\". Unfortunately, I can\\'t provide the value of `magic_function(3)` at this moment.\\n\\nIf you have any other questions or need assistance with something else, please let me know!' response_metadata={'token_usage': {'completion_tokens': 58, 'prompt_tokens': 195, 'total_tokens': 253}, 'model_name': 'gpt-4o', 'system_fingerprint': 'fp_729ea513f7', 'finish_reason': 'stop', 'logprobs': None} id='run-bb68d7ca-da76-43ad-80ab-23737a70c391-0'\n",
"{'input': 'what is the value of magic_function(3)?', 'output': 'Agent stopped due to max iterations.'}\n"
"content='Sorry, there was an error. Please try again.' name='magic_function' id='1a08b883-9c7b-4969-9e9b-67ce64cdcb5f' tool_call_id='call_pFdKcCu5taDTtOOfX14vEDRp'\n",
"content='It seems there was an error when trying to apply the magic function. Let me try again.' additional_kwargs={'tool_calls': [{'id': 'call_DA0lpDIkBFg2GHy4WsEcZG4K', 'function': {'arguments': '{\"input\":\"3\"}', 'name': 'magic_function'}, 'type': 'function'}]} response_metadata={'token_usage': {'completion_tokens': 34, 'prompt_tokens': 97, 'total_tokens': 131}, 'model_name': 'gpt-4o', 'system_fingerprint': 'fp_729ea513f7', 'finish_reason': 'tool_calls', 'logprobs': None} id='run-d571b774-0ea3-4e35-8b7d-f32932c3f3cc-0' tool_calls=[{'name': 'magic_function', 'args': {'input': '3'}, 'id': 'call_DA0lpDIkBFg2GHy4WsEcZG4K'}]\n",
"content='Sorry, there was an error. Please try again.' name='magic_function' id='0b45787b-c82a-487f-9a5a-de129c30460f' tool_call_id='call_DA0lpDIkBFg2GHy4WsEcZG4K'\n",
"content='It appears that there is a consistent issue when trying to apply the magic function to the input \"3.\" This could be due to various reasons, such as the input not being in the correct format or an internal error.\\n\\nIf you have any other questions or if there\\'s something else you\\'d like to try, please let me know!' response_metadata={'token_usage': {'completion_tokens': 66, 'prompt_tokens': 153, 'total_tokens': 219}, 'model_name': 'gpt-4o', 'system_fingerprint': 'fp_729ea513f7', 'finish_reason': 'stop', 'logprobs': None} id='run-50a962e6-21b7-4327-8dea-8e2304062627-0'\n"
]
}
],
@ -522,7 +787,7 @@
},
{
"cell_type": "code",
"execution_count": 30,
"execution_count": 17,
"id": "4b8498fc-a7af-4164-a401-d8714f082306",
"metadata": {},
"outputs": [
@ -549,7 +814,7 @@
" 'output': 'Agent stopped due to max iterations.'}"
"{'input': 'what is the value of magic_function(3)?', 'output': 'Agent stopped due to max iterations.'}\n"
]
@ -624,12 +889,12 @@
"id": "32a9db70",
"metadata": {},
"source": [
"The other way to set a max timeout is just via python's stdlib [asyncio](https://docs.python.org/3/library/asyncio.html)."
"The other way to set a single max timeout for an entire run is to directly use the python stdlib [asyncio](https://docs.python.org/3/library/asyncio.html) library."
"{'action': {'messages': [ToolMessage(content='Sorry, there was an error. Please try again.', name='magic_function', id='596baf13-de35-4a4f-8b78-475b387a1f40', tool_call_id='call_cR1oJuYcNrOmcaaIRRvh5dSr')]}}\n",
"{'input': 'what is the value of magic_function(3)?', 'output': 'Task Cancelled.'}\n"
"Task Cancelled.\n"
]
}
],
@ -665,6 +928,290 @@
"except TimeoutError:\n",
" print(\"Task Cancelled.\")"
]
},
{
"cell_type": "markdown",
"id": "4884ac87",
"metadata": {},
"source": [
"## `early_stopping_method`\n",
"\n",
"With LangChain's [AgentExecutor](https://api.python.langchain.com/en/latest/agents/langchain.agents.agent.AgentExecutor.html#langchain.agents.agent.AgentExecutor.iter), you could configure an [early_stopping_method](https://api.python.langchain.com/en/latest/agents/langchain.agents.agent.AgentExecutor.html#langchain.agents.agent.AgentExecutor.early_stopping_method) to either return a string saying \"Agent stopped due to iteration limit or time limit.\" (`\"force\"`) or prompt the LLM a final time to respond (`\"generate\"`)."
"content='Sorry there was an error, please try again.' name='magic_function' id='00d5386f-eb23-4628-9a29-d9ce6a7098cc' tool_call_id='call_bTURmOn9C8zslmn0kMFeykIn'\n",
" print({\"input\": query, \"output\": \"Agent stopped due to max iterations.\"})"
]
},
{
"cell_type": "markdown",
"id": "017fe20e",
"metadata": {},
"source": [
"## `trim_intermediate_steps`\n",
"\n",
"With LangChain's [AgentExecutor](https://api.python.langchain.com/en/latest/agents/langchain.agents.agent.AgentExecutor.html#langchain.agents.agent.AgentExecutor), you could trim the intermediate steps of long-running agents using [trim_intermediate_steps](https://api.python.langchain.com/en/latest/agents/langchain.agents.agent.AgentExecutor.html#langchain.agents.agent.AgentExecutor.trim_intermediate_steps), which is either an integer (indicating the agent should keep the last N steps) or a custom function.\n",
"\n",
"For instance, we could trim the value so the agent only sees the most recent intermediate step."
]
},
{
"cell_type": "code",
"execution_count": 22,
"id": "b94bb169",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Call number: 1\n",
"Call number: 2\n",
"Call number: 3\n",
"Call number: 4\n",
"Call number: 5\n",
"Call number: 6\n",
"Call number: 7\n",
"Call number: 8\n",
"Call number: 9\n",
"Call number: 10\n",
"Call number: 11\n",
"Call number: 12\n",
"Call number: 13\n",
"Call number: 14\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Stopping agent prematurely due to triggering stop condition\n"
"query = \"Call the magic function 4 times in sequence with the value 3. You cannot call it multiple times at once.\"\n",
"\n",
"for step in agent_executor.stream({\"input\": query}):\n",
" pass"
]
},
{
"cell_type": "markdown",
"id": "3d450c5a",
"metadata": {},
"source": [
"#### In LangGraph\n",
"\n",
"We can use the [`messages_modifier`](https://langchain-ai.github.io/langgraph/reference/prebuilt/#create_react_agent) just as before when passing in [prompt templates](#prompt-templates)."
]
},
{
"cell_type": "code",
"execution_count": 23,
"id": "b309ba9a",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Call number: 1\n",
"Call number: 2\n",
"Call number: 3\n",
"Call number: 4\n",
"Call number: 5\n",
"Call number: 6\n",
"Call number: 7\n",
"Call number: 8\n",
"Call number: 9\n",
"Call number: 10\n",
"Call number: 11\n",
"Call number: 12\n",
"Stopping agent prematurely due to triggering stop condition\n"