3fi N d dl mZ d dlmZ d dlmZ d dlmZmZ d dl m
Z
d dl m Z d dl
mZ d dlmZmZ d d lmZmZ d d
lmZ d d lmZmZ d d lmZ d d
l m Z d dl m Z d dl m!Z! d dl"m#Z# d dl$m%Z% eddd G d de Z&efddde dee d ed ed e'e(e) z d ef d Z*y!)" )Sequence)Any)
deprecated) AgentAction AgentFinish) Callbacks)BaseLanguageModel)BasePromptTemplate)AIMessagePromptTemplateChatPromptTemplate)RunnableRunnablePassthrough)BaseTool)
ToolsRendererrender_text_description)override)BaseSingleActionAgent
format_xmlXMLAgentOutputParser)agent_instructions)LLMChainz0.1.0create_xml_agentz1.0) alternativeremovalc e Zd ZU dZee ed< eed< ee dee
fd Z e de
fd Ze defd Ze dd eeee
f d
ed edeez fd Ze dd eeee
f d
ed edeez fd
Zy)XMLAgentaC Agent that uses XML tags.
Args:
tools: list of tools the agent can choose from
llm_chain: The LLMChain to call to predict the next action
Examples:
```python
from langchain_classic.agents import XMLAgent
from langchain
tools = ...
model =
```
tools llm_chainreturnc dgS )Ninput )selfs _/var/www/auto_recruiter/arenv/lib/python3.12/site-packages/langchain_classic/agents/xml/base.py
input_keyszXMLAgent.input_keys. s y c d t j t } | t j d z S )z,Return the default prompt for the XML agent.z{intermediate_steps})r
from_templater r ) base_prompts r& get_default_promptzXMLAgent.get_default_prompt3 s3 )667IJ 4 B B "
r( c t S )z Return an XMLAgentOutputParser.r r$ r( r& get_default_output_parserz"XMLAgent.get_default_output_parser; s
$%%r( Nintermediate_steps callbackskwargsc 4 d}|D ]( \ }}|d|j d|j d| dz
}* d}| j D ]! }||j d|j dz
}# |||d d d
gd } | j | | }
|
| j
j S
N zzzzz:
r#
z)r/ r questionstop)r0 )tool
tool_inputr name descriptionr
output_key r% r/ r0 r1 logaction observationr r9 inputsresponses r& planz
XMLAgent.plan@ s #5 FK
%89J9J8K,[M I
C
JJ :D {"T%5%5$6b9 9E : #& w$&78
>>&I>>1122r( c d K d}|D ]( \ }}|d|j d|j d| dz
}* d}| j D ]! }||j d|j dz
}# |||d d d
gd } | j
j
| | d { }
|
| j
j S 7 wr3 )r9 r: r r; r< r acallr= r> s r& aplanzXMLAgent.aplanY s #5 FK
%89J9J8K,[M I
C
JJ :D {"T%5%5$6b9 9E : #& w$&78
--f -JJ1122 Ks BB0B. B0)N)__name__
__module__ __qualname____doc__listr __annotations__r propertyr strr' staticmethodr r, r r. tupler r r r rD rG r$ r( r& r r s" " >1)
DI
2
&'; & & $3 {C'7!8 93 3 3
{ " 3 30 $3 {C'7!8 93 3 3
{ " 3 3r( r T)
stop_sequencellmr prompttools_rendererrR r! c ^ ddhj |j t |j z }|rd| }t | |j |t | }|r |du rdgn|}| j
| }n| }t j d |z |z t z S )
a Create an agent that uses XML to format its logic.
Args:
llm: LLM to use as the agent.
tools: Tools this agent has access to.
prompt: The prompt to use, must have input keys
`tools`: contains descriptions for each tool.
`agent_scratchpad`: contains previous agent actions and tool outputs.
tools_renderer: This controls how the tools are converted into a string and
then passed into the LLM.
stop_sequence: bool or list of str.
If `True`, adds a stop token of "" to avoid hallucinates.
If `False`, does not add a stop token.
If a list of str, uses the provided list as the stop tokens.
You may to set this to False if the LLM you are using
does not support stop sequences.
Returns:
A Runnable sequence representing an agent. It takes as input all the same input
variables as the prompt passed in does. It returns as output either an
AgentAction or AgentFinish.
Example:
```python
from langchain_classic import hub
from langchain_anthropic import ChatAnthropic
from langchain_classic.agents import AgentExecutor, create_xml_agent
prompt = hub.pull("hwchase17/xml-agent-convo")
model = ChatAnthropic(model="claude-3-haiku-20240307")
tools = ...
agent = create_xml_agent(model, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools)
agent_executor.invoke({"input": "hi"})
# Use with chat history
from langchain_core.messages import AIMessage, HumanMessage
agent_executor.invoke(
{
"input": "what's my name?",
# Notice that chat_history is a string
# since this prompt is aimed at LLMs, not chat models
"chat_history": "Human: My name is Bob\nAI: Hello Bob!",
}
)
```
Prompt:
The prompt must have input keys:
* `tools`: contains descriptions for each tool.
* `agent_scratchpad`: contains previous agent actions and tool outputs as
an XML string.
Here's an example:
```python
from langchain_core.prompts import PromptTemplate
template = '''You are a helpful assistant. Help the user answer any questions.
You have access to the following tools:
{tools}
In order to use a tool, you can use and tags. You will then get back a response in the form
For example, if you have a tool called 'search' that could run a google search, in order to search for the weather in SF you would respond:
searchweather in SF
64 degrees
When you are done, respond with a final answer between . For example:
The weather in SF is 64 degrees
Begin!
Previous Conversation:
{chat_history}
Question: {input}
{agent_scratchpad}'''
prompt = PromptTemplate.from_template(template)
```
r agent_scratchpadz#Prompt missing required variables: )r Tr6 )r8 c t | d S )Nr/ r )xs r& z"create_xml_agent.. s z!4H2I'J r( )rW )
differenceinput_variablesrL partial_variables
ValueErrorpartialbindr assignr ) rS r rT rU rR missing_varsmsgr8
llm_with_stops r& r r s s B /0;; f&>&>!??L 3L>Bo
^^ T%[) F $1T$9 }d+
"" J
!r( N)+collections.abcr typingr langchain_core._apir langchain_core.agentsr r langchain_core.callbacksr langchain_core.language_modelsr langchain_core.prompts.baser
langchain_core.prompts.chatr r langchain_core.runnablesr
r langchain_core.toolsr langchain_core.tools.renderr r typing_extensionsr langchain_classic.agents.agentr *langchain_classic.agents.format_scratchpadr 'langchain_classic.agents.output_parsersr #langchain_classic.agents.xml.promptr langchain_classic.chains.llmr r boolrL rO r r$ r( r& rw s $ * : . < : S B ) N & @ A H B 1 G!3UCY3$ Y3 DY3@ %< y '+
y y H y
y " y $s)#
y yr(