A programming framework for agentic AI
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Serializing Components \n",
"\n",
"AutoGen provides a {py:class}`~autogen_core.Component` configuration class that defines behaviours to serialize/deserialize component into declarative specifications. We can accomplish this by calling `.dump_component()` and `.load_component()` respectively. This is useful for debugging, visualizing, and even for sharing your work with others. In this notebook, we will demonstrate how to serialize multiple components to a declarative specification like a JSON file. \n",
"\n",
"\n",
"```{warning}\n",
"\n",
"ONLY LOAD COMPONENTS FROM TRUSTED SOURCES.\n",
"\n",
"With serilized components, each component implements the logic for how it is serialized and deserialized - i.e., how the declarative specification is generated and how it is converted back to an object. \n",
"\n",
"In some cases, creating an object may include executing code (e.g., a serialized function). ONLY LOAD COMPONENTS FROM TRUSTED SOURCES. \n",
" \n",
"```\n",
"\n",
"```{note}\n",
"`selector_func` is not serializable and will be ignored during serialization and deserialization process.\n",
"```\n",
"\n",
" \n",
"### Termination Condition Example \n",
"\n",
"In the example below, we will define termination conditions (a part of an agent team) in python, export this to a dictionary/json and also demonstrate how the termination condition object can be loaded from the dictionary/json. \n",
" "
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Config: {\"provider\":\"autogen_agentchat.base.OrTerminationCondition\",\"component_type\":\"termination\",\"version\":1,\"component_version\":1,\"description\":null,\"config\":{\"conditions\":[{\"provider\":\"autogen_agentchat.conditions.MaxMessageTermination\",\"component_type\":\"termination\",\"version\":1,\"component_version\":1,\"config\":{\"max_messages\":5}},{\"provider\":\"autogen_agentchat.conditions.StopMessageTermination\",\"component_type\":\"termination\",\"version\":1,\"component_version\":1,\"config\":{}}]}}\n"
]
}
],
"source": [
"from autogen_agentchat.conditions import MaxMessageTermination, StopMessageTermination\n",
"\n",
"max_termination = MaxMessageTermination(5)\n",
"stop_termination = StopMessageTermination()\n",
"\n",
"or_termination = max_termination | stop_termination\n",
"\n",
"or_term_config = or_termination.dump_component()\n",
"print(\"Config: \", or_term_config.model_dump_json())\n",
"\n",
"new_or_termination = or_termination.load_component(or_term_config)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Agent Example \n",
"\n",
"In the example below, we will define an agent in python, export this to a dictionary/json and also demonstrate how the agent object can be loaded from the dictionary/json."
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"from autogen_agentchat.agents import AssistantAgent, UserProxyAgent\n",
"from autogen_ext.models.openai import OpenAIChatCompletionClient\n",
"\n",
"# Create an agent that uses the OpenAI GPT-4o model.\n",
"model_client = OpenAIChatCompletionClient(\n",
" model=\"gpt-4o\",\n",
" # api_key=\"YOUR_API_KEY\",\n",
")\n",
"agent = AssistantAgent(\n",
" name=\"assistant\",\n",
" model_client=model_client,\n",
" handoffs=[\"flights_refunder\", \"user\"],\n",
" # tools=[], # serializing tools is not yet supported\n",
" system_message=\"Use tools to solve tasks.\",\n",
")\n",
"user_proxy = UserProxyAgent(name=\"user\")"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{\"provider\":\"autogen_agentchat.agents.UserProxyAgent\",\"component_type\":\"agent\",\"version\":1,\"component_version\":1,\"description\":null,\"config\":{\"name\":\"user\",\"description\":\"A human user\"}}\n"
]
}
],
"source": [
"user_proxy_config = user_proxy.dump_component() # dump component\n",
"print(user_proxy_config.model_dump_json())\n",
"up_new = user_proxy.load_component(user_proxy_config) # load component"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{\"provider\":\"autogen_agentchat.agents.AssistantAgent\",\"component_type\":\"agent\",\"version\":1,\"component_version\":1,\"description\":null,\"config\":{\"name\":\"assistant\",\"model_client\":{\"provider\":\"autogen_ext.models.openai.OpenAIChatCompletionClient\",\"component_type\":\"model\",\"version\":1,\"component_version\":1,\"config\":{\"model\":\"gpt-4o\"}},\"handoffs\":[{\"target\":\"flights_refunder\",\"description\":\"Handoff to flights_refunder.\",\"name\":\"transfer_to_flights_refunder\",\"message\":\"Transferred to flights_refunder, adopting the role of flights_refunder immediately.\"},{\"target\":\"user\",\"description\":\"Handoff to user.\",\"name\":\"transfer_to_user\",\"message\":\"Transferred to user, adopting the role of user immediately.\"}],\"model_context\":{\"provider\":\"autogen_core.model_context.UnboundedChatCompletionContext\",\"component_type\":\"chat_completion_context\",\"version\":1,\"component_version\":1,\"config\":{}},\"description\":\"An agent that provides assistance with ability to use tools.\",\"system_message\":\"Use tools to solve tasks.\",\"reflect_on_tool_use\":false,\"tool_call_summary_format\":\"{result}\"}}\n"
]
}
],
"source": [
"agent_config = agent.dump_component() # dump component\n",
"print(agent_config.model_dump_json())\n",
"agent_new = agent.load_component(agent_config) # load component"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"A similar approach can be used to serialize the `MultiModalWebSurfer` agent.\n",
"\n",
"```python\n",
"from autogen_ext.agents.web_surfer import MultimodalWebSurfer\n",
"\n",
"agent = MultimodalWebSurfer(\n",
" name=\"web_surfer\",\n",
" model_client=model_client,\n",
" headless=False,\n",
")\n",
"\n",
"web_surfer_config = agent.dump_component() # dump component\n",
"print(web_surfer_config.model_dump_json())\n",
"\n",
"```"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Team Example\n",
"\n",
"In the example below, we will define a team in python, export this to a dictionary/json and also demonstrate how the team object can be loaded from the dictionary/json."
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{\"provider\":\"autogen_agentchat.teams.RoundRobinGroupChat\",\"component_type\":\"team\",\"version\":1,\"component_version\":1,\"description\":null,\"config\":{\"participants\":[{\"provider\":\"autogen_agentchat.agents.AssistantAgent\",\"component_type\":\"agent\",\"version\":1,\"component_version\":1,\"config\":{\"name\":\"assistant\",\"model_client\":{\"provider\":\"autogen_ext.models.openai.OpenAIChatCompletionClient\",\"component_type\":\"model\",\"version\":1,\"component_version\":1,\"config\":{\"model\":\"gpt-4o\"}},\"handoffs\":[{\"target\":\"flights_refunder\",\"description\":\"Handoff to flights_refunder.\",\"name\":\"transfer_to_flights_refunder\",\"message\":\"Transferred to flights_refunder, adopting the role of flights_refunder immediately.\"},{\"target\":\"user\",\"description\":\"Handoff to user.\",\"name\":\"transfer_to_user\",\"message\":\"Transferred to user, adopting the role of user immediately.\"}],\"model_context\":{\"provider\":\"autogen_core.model_context.UnboundedChatCompletionContext\",\"component_type\":\"chat_completion_context\",\"version\":1,\"component_version\":1,\"config\":{}},\"description\":\"An agent that provides assistance with ability to use tools.\",\"system_message\":\"Use tools to solve tasks.\",\"reflect_on_tool_use\":false,\"tool_call_summary_format\":\"{result}\"}}],\"termination_condition\":{\"provider\":\"autogen_agentchat.conditions.MaxMessageTermination\",\"component_type\":\"termination\",\"version\":1,\"component_version\":1,\"config\":{\"max_messages\":2}}}}\n"
]
}
],
"source": [
"from autogen_agentchat.agents import AssistantAgent, UserProxyAgent\n",
"from autogen_agentchat.conditions import MaxMessageTermination\n",
"from autogen_agentchat.teams import RoundRobinGroupChat\n",
"from autogen_ext.models.openai import OpenAIChatCompletionClient\n",
"\n",
"# Create an agent that uses the OpenAI GPT-4o model.\n",
"model_client = OpenAIChatCompletionClient(\n",
" model=\"gpt-4o\",\n",
" # api_key=\"YOUR_API_KEY\",\n",
")\n",
"agent = AssistantAgent(\n",
" name=\"assistant\",\n",
" model_client=model_client,\n",
" handoffs=[\"flights_refunder\", \"user\"],\n",
" # tools=[], # serializing tools is not yet supported\n",
" system_message=\"Use tools to solve tasks.\",\n",
")\n",
"\n",
"team = RoundRobinGroupChat(participants=[agent], termination_condition=MaxMessageTermination(2))\n",
"\n",
"team_config = team.dump_component() # dump component\n",
"print(team_config.model_dump_json())\n",
"\n",
"await model_client.close()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": ".venv",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.11.9"
}
},
"nbformat": 4,
"nbformat_minor": 2
}