react_test.py

from langchain.agents import load_tools
from langchain.agents import initialize_agent
from langchain.agents import AgentType, ZeroShotAgent, AgentExecutor
from langchain.llms import OpenAI
from model_context import get_watsonx_predictor
from langchain.llms import Ollama
from langchain.llms import HuggingFaceTextGenInference

import langchain

#langchain.debug = True

##### temporary tools
from langchain.tools.base import BaseTool


class YAMLGeneratorTool(BaseTool):
    """Tool for generating kubernetes and openshift yaml"""

    name = "yaml_generator_tool"
    description = "Useful for generating YAML for kubernetes and openshift"

    def _run(self, tool_input: str) -> str:
        """Run the tool."""
        return '''apiVersion: "autoscaling.openshift.io/v1"
kind: "ClusterAutoscaler"
metadata:
  name: "default"
spec:
  resourceLimits:
    maxNodesTotal: 24 
'''

class DocsSummaryTool(BaseTool):
    """Tool for retrieving documentation summaries from the RAG"""

    name = "docs_summary_tool"
    description = "Documentation useful for determining what steps are required to perform an OpenShift or Kubernetes task"

    def _run(self, tool_input: str) -> str:
        """Run the tool."""
        return "1. Create a ClusterAutoscaler YAML. The cluster maximum size is required."


####

#### model setup
# llm = get_watsonx_predictor(model="tiiuae/falcon-180b", min_new_tokens=5, verbose=True)
llm = get_watsonx_predictor(
    model="codellama/codellama-34b-instruct", min_new_tokens=5, verbose=True
)
# llm = Ollama(
#    model="mistral:7b-instruct",
#    base_url="http://ollama-rcs5.apps.rhods-internal.61tk.p1.openshiftapps.com",
#    verbose=True,
# 		temperature=0.05
# )
# llm = OpenAI(temperature=0, verbose=True)
# llm = HuggingFaceTextGenInference(
#    inference_server_url="http://tgis-deployment-rcs5.apps.rhods-internal.61tk.p1.openshiftapps.com",
#    max_new_tokens=200,
#    temperature=0.01,
#    repetition_penalty=1.03,
# )
####

#### tool setup
# tools = load_tools(["yaml_generator_tool", "llm-math"], llm=llm)
tools = [DocsSummaryTool(),YAMLGeneratorTool()]

agent_executor = initialize_agent(
    tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True
)
agent_executor.invoke(
    {"input": "How can I configure my cluster for autoscaling up to 10 nodes?"}
)


####

# prefix = """
# [INST]
# Answer the following questions as best you can. You have access to the following tools:
#
# """
#
# format_instructions = """
# Use the following format:
#
# Thought: you should always think about what to do
# Action: the action to take, should be one of [{tool_names}]
# Action Input: the input to the action
# Observation: the result of the action
# ... (this Thought/Action/Action Input/Observation can repeat N times)
# Thought: I now know the final answer
# Final Answer: the final answer to the original input question
# [/INST]
#
# """
#
# suffix = """
# Question: {input}
# Thought:{agent_scratchpad}
# """

# agent = ZeroShotAgent.from_llm_and_tools(llm=llm, tools=tools, prefix=prefix, format_instructions=format_instructions, suffix=suffix, verbose=True)
# agent_executor = AgentExecutor.from_agent_and_tools(agent=agent, tools=tools, verbose=True)