Agents#
Note
Some applications require not just a predetermined chain of calls to LLMs/other tools, but potentially an unknown chain that depends on the user’s input. In these types of chains, there is an agent which has access to a suite of tools. Depending on the user input, the agent can then decide which, if any, of these tools to call.
At the moment, there are two main types of agents:
Action Agents: these agents decide the actions to take and execute that actions one action at a time.
Plan-and-Execute Agents: these agents first decide a plan of actions to take, and then execute those actions one at a time.
When should you use each one? Action Agents are more conventional, and good for small tasks. For more complex or long running tasks, the initial planning step helps to maintain long term objectives and focus. However, that comes at the expense of generally more calls and higher latency. These two agents are also not mutually exclusive - in fact, it is often best to have an Action Agent be in charge of the execution for the Plan and Execute agent.
Action Agents#
High level pseudocode of the Action Agents:
The user input is received
The agent decides which tool - if any - to use, and what the tool input should be
That tool is then called with the tool input, and an observation is recorded (the output of this calling)
That history of tool, tool input, and observation is passed back into the agent, and it decides the next step
This is repeated until the agent decides it no longer needs to use a tool, and then it responds directly to the user.
The different abstractions involved in agents are:
Agent: this is where the logic of the application lives. Agents expose an interface that takes in user input along with a list of previous steps the agent has taken, and returns either an AgentAction or AgentFinish
AgentAction corresponds to the tool to use and the input to that tool
AgentFinish means the agent is done, and has information around what to return to the user
Tools: these are the actions an agent can take. What tools you give an agent highly depend on what you want the agent to do
Toolkits: these are groups of tools designed for a specific use case. For example, in order for an agent to interact with a SQL database in the best way it may need access to one tool to execute queries and another tool to inspect tables.
Agent Executor: this wraps an agent and a list of tools. This is responsible for the loop of running the agent iteratively until the stopping criteria is met.
Getting Started: An overview of agents. It covers how to use all things related to agents in an end-to-end manner.
Agent Construction:
Although an agent can be constructed in many way, the typical way to construct an agent is with:
PromptTemplate: this is responsible for taking the user input and previous steps and constructing a prompt to send to the language model
Language Model: this takes the prompt constructed by the PromptTemplate and returns some output
Output Parser: this takes the output of the Language Model and parses it into an AgentAction or AgentFinish object.
Additional Documentation:
Tools: Different types of tools LangChain supports natively. We also cover how to add your own tools.
Agents: Different types of agents LangChain supports natively. We also cover how to modify and create your own agents.
Toolkits: Various toolkits that LangChain supports out of the box, and how to create an agent from them.
Agent Executor: The Agent Executor class, which is responsible for calling the agent and tools in a loop. We go over different ways to customize this, and options you can use for more control.
Plan-and-Execute Agents#
High level pseudocode of the Plan-and-Execute Agents:
The user input is received
The planner lists out the steps to take
The executor goes through the list of steps, executing them
The most typical implementation is to have the planner be a language model, and the executor be an action agent.